Colloids and Surfaces A: Physicochemical and Engineering Aspects, 84 (1994)141-193 141

Elsevier Science B.V.

Review

Techniques for measuring the composition (oil and

water content) of emulsions - a state of the art review

Rajinder Pal
Depavtlnent of Chemical Engineering, University ofWaterloo, Waterloo, Ont. N2L 3G1, Canada

(Received 14 August 1993; accepted 12 November 1993)

Abstract

Measurement of the oil and water content of emulsions is important industrially. However, despite its importance,
little attention has been given to this topic in colloid journals. The literature available on emulsion composition
measurement is scattered in various journals, reports and patents. This is the first comprehensive review of the topic.
Following a brief discussion about sampling and associated problems, the working principles of various composition
measurement techniques available to date are discussed, albeit briefly. Where possible, the limitations of the techniques
are pointed out. Although the review emphasizes the petroleum industry, it has relevance to the field of emulsions
in general.
The techniques available are classified into two broad groups: destructive techniques and non-destructive techniques,
depending upon whether or not the emulsion undergoes any permanent degradation during the measurement. The
destructive techniques discussed are titration and related methods, total organic carbon analysis, centrifugation,
distillation, flash vaporization, solvent extraction and differential scanning calorimetry. The non-destructive techniques
discussed are electrical methods (capacitance/conductance techniques), microwave-based techniques, radiation scattering
techniques (turbidity measurement, neutron scattering, depolarization measurement), spectroscopic techniques (IR
absorption, UV absorption, NMR), ultrasonic techniques (sound scattering, use of the acoustic velocitimeter,
measurement of the ultrasonic vibration potential), density-based techniques (use of a radiation gauge, vibrating
densitometers, a method involving continuous weighing, pressure drop measurement) and viscosity-based techniques.
One major drawback of the current literature is the general lack of experimental data available on various techniques.
Only a few techniques have been investigated in detail; in several cases, little or no experimental work is reported to
test the full range of applicability of these techniques. Thus it is difficult to draw any detailed comparisons between the
various techniques.
It is hoped that this review will stimulate further interest in this important area of research. It is recommended that
future research should focus on experimental work so as to determine the full range of the applicability and
limitations/accuracy of the various available techniques.

Key words: Composition measurement; Emulsions; Oil content; Techniques; Water content

1. Introduction these droplets are suspended is called the “con-

tinuous phase”. The continuous phase is sometimes
1 .I. Dqjinition and classification of emulsions referred to as the external phase, dispersion
medium or suspending medium. The droplets of
An “emulsion” is a two-phase oil/water system emulsions usually exceed 0.5 ym in diameter so
where one of the phases is dispersed as droplets in that they are visible under the optical microscope.
the other. The phase which is present in the form Emulsions also contain a third component, called
of droplets is referred to as the “dispersed phase” the emulsifying agent or emulsifier, which has two
and the phase which forms the matrix in which principal functions: (1) to decrease the interfacial

SSDI 0927-7757(93)02711-M
tension between oil and water, thereby enabling the surface can also be instrumental in producing
easier formation of the emulsion; (2) to stabilize emulsions because of the high shear. The surface
the dispersed phase against coalescence once it facilities such as valves, chokes and turns also
is formed. favour emulsion formation. The source of emulsify-
ing agents in the produced emulsions is the crude
1 .I .I. Clmssijcntion qf emulsions oil itself, which contains a variety of surface-active
Emulsions can be classified into three broad agents such as polycyclics and long-chain fatty
groups: acids [ 2,3]. Finely divided solids such as clay and
(a) water-in-oil emulsions, designated as “W/O sand are also present in the crude oil and these
emulsions”; are well-known stabilizing agents for emulsions.
(b) oil-in-water emulsions, designated as “O/W In primary oil production operations, where the
emulsions”; reservoir’s natural energy (pressure) is used to
(c) multiple emulsions. produce the oil, W/O emulsions are more com-
The W,/O emulsions consist of water droplets monly encountered. The appearance of O/W emul-
dispersed in a continuum of oil. It is this type of sions occurs usually when a given oii field becomes
emulsion that is more commonly encountered in old and produces a greatly increased amount of
the petroleum industry and these emulsions are, water [4].
therefore, referred to as “regular emulsions”. The In enhanced oil recovery operations, a fluid is
O/W type of emulsion consists of oil droplets often injected into the formation to push a front
dispersed in water. In the petroleum industry, the of oil to a producing well. A variety of fluids are
O/W emulsions are called “reverse emulsions”, used for this purpose, such as water, aqueous
being the reverse of the better known W/O types. polymer solution, caustic solution, aqueous surfac-
Multiple emulsions consist of multiple droplets; for tant solutions, microemulsion and steam. Most of
example. a W/O/W multiple emulsion consists of these fluids not only push the oil from the forma-
one or more small water droplets entrapped within tion into the producing well but they also help
large oil droplets, which, in turn, are suspended in recover the oil adhered to the walls of the pores
a continuum of water. The multiple emulsions are (due to strong capillary forces). Very often. how-
not as common as the simple binary emulsions. ever, channelling and breakthrough of fluid into
the producing well occurs, and as a result, emul-
1.2. Applications of’endsions sions are formed [3]. Both 0,/W and W/O emul-
sions are produced.
Emulsions are encountered at nearly every step In many oil-production-related operations,
of the petroleum production and recovery opera- emulsions are deliberately produced to take advan-
tions, viz. within the underground porous media tage of their unique physical properties. For exam-
(oil reservoir), at wellheads, in phase separators, in ple, large quantities of emulsion-based fluids are
flotation units, in crude oil transport facilities, and used for oil-well drilling [S], as well as for well-
at various stages of the refining process. fracturing and acidizing purposes [6]. In situations
As crude oil rises from the underground oil where the oil production rate from a given oil well
reservoir. its passage through narrow openings in is severely limited because of the high viscosity of
the porous matrix, accompanied by water, gives the crude oil produced, an aqueous surfactant
rise to conditions favourable to the formation of solution is often injected into the production-well
emulsions. The emulsification of oil and water is bore to convert the high viscosity crude oil into a
believed to occur mostly in the formation near the low viscosity O/W emulsion [ 7- 121. The reduction
well bore where the velocity gradients are very in viscosity of the produced fluids enhances the oil
high [ 11. The pump used to lift the well fluids to production rate quite substantially owing to an
R. Pal/Col/oids SurfacesA: Physicochem. Eng. Aspects 84 (I 994) 141-l 93 143

increase in the rod dropping rate and a decrease tile, bitumen, paints, lubricants, polymer and
in the flow line pressure loss. This process whereby transport industries [22-391. Examples of non-
aqueous solutions of surface-active chemicals are petroleum emulsions are milk, mayonnaise, salad
pumped down into the producing well to enhance dressings, cake batters, meat emulsions, butter,
the oil production rate is referred to as “downhole margarine, several anti-bacterial and anti-allergic
emulsification”. creams, foundation and vanishing creams, sun-
The pipeline transportation of highly viscous screen lotions and creams, hair creams, shaving
crude oils in the form of O/W emulsions has creams, hand lotions and creams, cold creams,
received special attention in recent years insecticides for plants and animals, emulsified
[ 11,133181. The transportation of extremely vis- explosives, polishes of oil or wax, fat liquors
cous crude oils, such as bitumen and heavy oils, employed in the treatment of leather, anti-static
by pipelines is difficult, especially during cold and waterproofing agents employed in the textile
weather. To facilitate the flow of very viscous crude industry, emulsified lubricants for fibres, bitumen
in pipelines, it is necessary to reduce its viscosity emulsions for the construction of roads and high-
either through the installation of heating equip- ways, emulsion paints, lubricants for drilling and
ment at frequent intervals along the pipeline or cutting tools, reaction systems in emulsion poly-
through the addition of a low viscosity hydro- merization processes, etc. Also, many hazardous
carbon diluent. The former expedient is expensive materials, such as alkali metals and explosives, are
and inconvenient, and the latter requires the avail- often handled in emulsion form.
ability of relatively large amounts of cheap diluent.
These problems, however, can be avoided if the 1.3. Importunce of composition measurement
viscous crudes are transported as concentrated
O/W emulsions. The viscosity of O/W emulsions Although the major application of composition
is, in general, very low compared with that of the measurement of emulsions is in the petroleum
crude oil. This method has already been utilized industry, a knowledge of the emulsion composition
commercially in a pipeline 21 km in length and is important in many other applications also,
20 cm in diameter in California and in a pipeline mainly because the bulk properties of the emul-
238 km in length and 51 cm in diameter in sions, such as emulsion rheology, are a strong
Indonesia [ 131. function of the composition. Specific applications,
Recently, interest has also been shown in the where measurement of emulsion composition is
use of emulsions as blocking agents to control the important are given below.
permeability of the formation [ 191. High viscosity (a) Crude oil being an expensive commodity,
emulsions are also being considered as displace- both fiscal authorities and operators involved in
ment fluids to displace oil in the enhanced oil oil production operations need to know the exact
recovery processes [ 20-211. amounts of crude oil that they produce, sell and
The applications of emulsions discussed above buy. Thus, accurate measurement of oil and water
relate to the petroleum industry. Emulsions play production rates is important to prevent paying
an important role in many other industries as well. for water rather than oil ~ an underestimation of
Numerous products of commercial importance are about 0.1% in the water content measurement can
sold in emulsion form. Books have been written result in the recipient of a typical oil tanker load
describing the commercial applications of emulsion paying over $25 000 for water instead of oil [40].
systems [ 22-251. The non-petroleum industries Also, in many situations, the fluids produced from
where emulsions are of considerable importance several different oil-wells in a given field are com-
includes food, medical and pharmaceutical, cosme- bined into a single stream before any treatment.
tics, agriculture, explosives, polishes, leather, tex- Quite often, the oil wells are owned by different
owners and, therefore, it is essential to monitor gars, etc. [ 421. Again, environmental regulations
accurately the production of each well for the require accurate monitoring of oil discharges.
purpose of royalty calculations. (d) In many applications involving large-scale
(b) For oil field management, the production mechanical equipment (ship engines and power
rates of oil and water from each well must be station turbines), on-line monitoring of the water
measured. For example, if excessive water is being content of the lubricating oil 140.431 flowing into
produced on an injection field (an oil field where the machine, is necessary to prevent malfunction
of the machine components. For example. if the
a secondary recovery method involving water injec-
water content of the lubricating oil flowing into a
tion is being used). appropriate action could be
ship engine exceeds a certain limit (say 0.5%). it
taken to avoid compounded losses due to pumping
may cause the seizure of bearings, leading to
the water down the injection well and then bringing
serious consequences [40].
it back up to the top in the production well.
The presence of dispersed or free water in jet
In many off-shore oil fields, the water content
aircraft fuels can cause severe hazards at extreme
of the fluids produced can be very high. This is
altitudes where water may form ice-crystals leading
especially true for marginal fields and fields nearing to blockage of fuel line. orifices and screens in the
the ends of their lives. In such cases, it is extremely fuel system 1441. Thus accurate monitoring of the
important to know the exact oil content of the water content of aviation fuels is important.
fluids produced so that proper decisions can be (e) In several applications where two liquid
made regarding whether or not production should immiscible phases are contacted, on-line measure-
be continued. ment of the dispersed-phase concentration is
(c) Large quantities of O/W emulsions, oily required for the analysis, control and safety of the
wastewaters, are generated in petroleum refineries. operation. For example, the key factor in the
The major contributors to generated wastewaters control and analysis of two-phase liquid/liquid
are desalters, crude and product storage tanks, column extractors is the ability to monitor continu-
slop oil tanks, etc. Environmental regulations ously the dispersed-phase volume fraction in an
require that the hydrocarbon concentration in all accurate manner [ 4.53.
waste streams be less than a certain limit (typically (f) In the food industry. measurement of the
10 ppm). This requires accurate monitoring of the water content of the food products (many products
hydrocarbon content of oily water discharges. In are in emulsion form) is often a legal requirement;
a recent study carried out by Agar et al. [41], it for example, there is a legal limit of 16% moisture
content for butter. This is to restrict food manufac-
is shown that the key factor in controlling the oil
turers from adding more than the required amount
discharge concentrations from the refining and
of water to the products. Another food industry
petrochemical industries is improved instrumenta-
application which involves accurate measurement
tion for monitoring the oil content of the emulsions.
of the composition of an emulsion is milk screening.
They recommend that the emulsion monitoring
It is important to monitor the fat content of milk
probes be installed at various stages of the process
so as to provide a basis for checking the acceptabil-
to control or even eliminate completely the oil
ity of the milk in terms of its freedom from
discharge. adulteration [ 461.
It should be noted further that the problem of
oily water discharges is not restricted to petroleum 2. Basic aspects of composition measurement
refineries. The O/W emulsions are found in waste-
water effluent streams from many other sources as There are two different approaches that are used
well, such as rolling mills, chemical processing and to measure the composition (oil and water content)
manufacturing plants. ships. washracks and han- of emulsions:
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 145

(a) direct measurement of composition using an - MAIN I INE

EMULSION
in-line emulsion analyzer;
(b) indirect measurement of composition using l-_--J
MIXER
DEVICE
sampling and subsequent laboratory analysis of
the collected sample. Fig. 2. “Main-line” installation of device.
The latter approach, which involves the collec-
tion of a representative sample of the emulsion repair/recalibrate/remove/reinstall the device with-
from the pipeline or any other process equipment, out interrupting the flow in the main pipeline.
and subsequent analysis of the sample in the Furthermore, if the sampler or direct device being
laboratory, is more commonly used in the oil used is intrusive in nature and it is not capable of
production operations although increasing use is withstanding continuous operation on a long-term
being made of direct methods. However, it should basis owing to fragile components and/or hostile
be noted that the accuracy of the indirect method flow conditions (presence of dirt, flow transients,
(sampling and subsequent laboratory analysis) is fouling, temperature variations), a by-pass
highly dependent on the extent to which the col- approach is recommended as the device can then
lected sample is representative of the pipeline com- be used intermittently on a day-to-day basis.
position. Even when the flow upstream of the In the main-line method, the device is installed
sampling port is homogeneous and the emulsion in the main pipeline (see Fig. 2). While the advan-
is well dispersed, the collected sample may not be tage of the main-line approach is that the composi-
representative of the actual composition in the tion passing down the device is the actual
pipeline. Thus special care is required to ensure composition, the disadvantage lies in the difficulties
representative sampling. Details of sampling and associated with removing/recalibrating/reinstalling
associated errors are discussed in the next section. the device without interrupting the main flow.
There are two basic methods for installing a Furthermore, the device, if an intrusive type, should
sampling device or a direct emulsion analyzer in be capable of withstanding the main flow condi-
the pipeline [47,48]. These are referred to as “by- tions on a long-term basis without undergoing
pass” and “main-line” methods respectively. In the substantial wear and tear.
by-pass method, the device is installed in a by-pass
line from the main flow as shown in Fig. 1. It is 2.1. Sumpling
obvious that the accuracy of the by-pass type
devices is highly dependent on the extent to which On-line samplers for the measurement of water
the flow in the by-pass line is representative of the and oil contents of emulsions are widely used in
composition in the main line. Therefore it is impor- the petroleum industry. The samplers used in the
tant that an in-line mixer be installed upstream oil industry are broadly classified into two groups:
of the flow division junction. The major advan- grab or internal samplers, and fast loop or external
tage of the by-pass approach is that one can samplers [47,49,50].
Grab samplers are devices which take small
MAIN LINE volumes of fluid from the line at discrete time
EMULSION intervals to build up a representative sample of the
emulsion over a period of time. The current
BY-PASS LOOP international practice in the petroleum industry is
to collect 10000 “grabs” of about 1 ml each, i.e. a
total volume of about 10 1, before the sample is
DEVICE
sent out for laboratory analysis. Quite often, the
Fig. 1. “By-pass” installation of device. sampling system is designed to provide propor-
146 R. Pul/Colloids Surfuces A: Physiwchrrn. Eng. Aspects 84 (1994) 141-l 93

tional-to-flow sampling [SO] and the signal from probes and the factors which can lead to significant
the flowmeter (installed in the pipeline for total sampling errors.
flow measurement) is used to set the sampling In any sampling method, the sampling probe
frequency (see Fig. 3). plays a key role. Therefore the design and orienta-
In the fast loop sampler, a very small fraction tion of the probe are important considerations in
of the flow is diverted to a collection vessel by a sampling operations. There are at least four
sampling probe inserted into the pipeline, i.e. the different designs available for the sampler probe
material is extracted continuously from the pipe- [49,51] (see Fig. 5): 45’ opening, pitot, circular
line, rather than at discrete intervals as in the case port, and NEL pitot probes; the pitot-type sam-
of grab samplers, until the desired amount of pling probes (L-shaped probes) are widely used
sample is collected (Fig. 4). both for emulsion sampling and solids slurry sam-
The main concern with the use of samplers is pling 151-531. The probes can be installed either
that the sample collected may not be exactly in a horizontal section of a pipe or in a vertical
representative of the actual composition inside the section, pointing either upstream or downstream.
pipeline. Even when the flow upstream of the The most popular sampling arrangement, however,
sampling probe is homogeneous, the composition consists of an L-shaped probe pointing upstream;
of the collected sample may differ from the actual both horizontal and vertical orientations are in
composition inside the pipeline. Thus, sampling common use.
can give erroneous results if it is not carried out The performance of any sampling device is eval-
properly. In the following paragraphs, we will uated in terms of sampling efficiency E, defined
discuss some of the commonly used sampling as [52].

E = &/$.A (1)
FLOW
METER
where &,, is the measured concentration (volume
basis) of the dispersed phase and $A is the actual
I , I concentration inside the pipeline. If the concen-
SAMPLING PROBE

.:I
j FLOW
SAMPLE
1SIGNAL__j-__________ EXTRACTION
SYSTEM
GRAB 45’ OPENING PROBE PITOT PROBE
FREOUENCY 1
CONTROLLER _._--_ ---\ 8
-
i
_<
45’

Fig. 3. Proportional-to-flow sampling arrangement

MIXER

CIRCULAR PORT PROBE NEL PITOT PROBE

SAMPLlNG 1
PROBE
\ THREE-WAY
VALVE
J
4

Fig. 4. Schematic diagram of a fast loop sampler Fig. 5. Different designs of the sampler probe.
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 147

ISOKINETIC SAMPLING
tration of dispersed droplets is not uniform across
the cross-section of the pipe, then dA is the local STREAM LINES

concentration upstream of the probe.

SAMPLING PROBE
The sampling efficiency may deviate from an
ideal value of unity owing to several factors
[49,51-551; these include droplet inertia, droplet
coalescence, droplet bouncing at the probe, and
the structure of the flow ahead of the sampler.
SUE-ISOKINETIC SAMPLING

2.1 .I. Droplet inertia STREAM LINES

The densities of oil and water are different, and

SAMPLING PROBE
therefore the inertia of the same volumes of oil
and water are different. This difference in inertia is
not a significant problem in the case of O/W
emulsions where oil droplets, having lower inertia,
can easily follow the changes in flow direction.
However, in the case of W/O emulsions, the water SUPER-ISOKINETIC SAMPLING

droplets, having higher inertia, do not readily STREAM LINES

follow the changes in flow direction, and conse-

SAMPLING PROBE
quently sampling errors can occur.
Figure 6 illustrates the effect of water droplet
inertia on the sampling performance of L-shaped
probes [ 51,521. The probe is assumed to be very
thin and of negligible wall thickness with its axis Fig. 6. Schematic illustration of the effect of water droplet
inertia on sampler performance [ 51,521.
coinciding with that of the pipe. Three situations
are considered: isokinetic sampling, where the
velocity of the fluid in the sampler probe (U) is content of the sample is significantly higher than
the same as that of the fluid upstream of the probe the actual composition in the pipeline.
(U,), i.e. U/U, = 1; subisokinetic sampling, where In the case of superisokinetic sampling, where
U/U, < 1; and superisokinetic sampling, where U > UO, one observes a lower concentration of
U/U,> 1. water in the collected sample compared with the
In the case of isokinetic sampling, where U = actual concentration inside the pipeline, due to
U,, the sample composition is the same as that in inertia effects.
the pipe. The difference between oil and water It should further be noted that if the water
inertia does not play any role. droplets are not uniform in size, i.e. there exists a
When the sampling velocity is less than the size distribution of droplets, then the size distribu-
pipeline velocity, i.e. we have subisokinetic sam- tion in the collected sample will be different from
pling where U < U,, a distortion of stream lines the actual distribution, inside the pipe, when the
occurs ahead of the sampler. While the droplets of sampling conditions are anisokinetic. In the case
lower inertia, say oil droplets in an O/W emulsion, of subisokinetic sampling, the concentration of
can follow the distorted fluid streamlines readily, small droplets in the sample will be lower because
droplets of higher inertia, say water droplets in a smaller droplets are able to follow the stream lines;
W/O emulsion, are unable to follow the stream in superisokinetic sampling, the concentration of
lines; instead they continue to travel in more or fine droplets in the sample will be higher than the
less straight lines, and consequently the water actual value.
 In anisokinetic sampling situations, the response no influence on the sampling efficiency, E = 1 from
of the water droplets (in W/O emulsions) to the above equation. For subisokinetic sampling,
changes in flow direction depends on the ratio of where U/U,< 1, the sampling efficiency is greater
the droplet inertia force to the viscous drag force than unity, i.e. the concentration is higher in the
acting on the droplet; this ratio is known as the sample, and it increases with an increase in the
Stokes number or Barth number k’, [52]: I<, value. For superisokinetic sampling, where
U/U, > 1, the sampling efficiency is less than unity,
K, = y,d,2U,/1 8ycR (2)
i.e. the concentration is less in the sample, and its
where pp is the density of the droplet, Lf, is the deviation from unity increases with the increase in
diameter of the droplet, U, is the upstream local E;, value. It should be noted that with an increase
velocity, qC is the fluid (continuous-phase) viscosity in the dispersed-phase concentration the drag force
and R is the radius of the sampling probe opening. on the particles increases but this is not accounted
The higher the K, value the greater is the for through I<,. In order to account for the concen-
deviation of sampling efficiency from unity. tration effect on the drag force, the more appro-
Figure 7 shows some of the results of Iinoya priate viscosity in the expression for K, (Eq. (2))
152,561 concerning the effect of inertia on the would be the emulsion viscosity rather than the
sampling efficiency for a very dilute dispersed continuous-phase viscosity.
system. The plots are calculated from the following Figure 8 gives some of the experimental results
empirical correlation proposed by Iinoya [56] for of Nasr-El-Din et al. [ 531 obtained for solid/liquid
L-shaped probes: slurries using an L-shaped probe, either thin walled
or tapered wall. For isokinetic sampling, the sam-
E=(O.031&+0.22) [(U&J)-l]+l (3)
pling efficiency is unity. For subisokinetic sampling,
For isokinetic sampling, where U/U,= 1, K, has the efficiency is greater than unity as expected; the
efficiency is less than unity for superisokinetic
100.00 sampling.

2.1.2. Droplet coalescence

King et al. [49] in their study of W/O samplers,
10.00 observed that if the sampling was done at low
sampling rates, that is, s~lbisokinetically with
a U <0.2U,, the water droplets exhibited a tendency
to coalesce ahead of the sampler probe.
q
1 .oo Consequently, when the pitot-type probes were
3

0.10 1

$ 1.4

2
1.0
0.01

Fig. 7. Effect of droplet inertia on the sampling efficiency:

Iinoga’s results [56]. Fig. 8. Experimental results of Nasr-El-Din et al. 1531.
R. PallColloids Surfaces A: Physicochrm. Eng. Aspects 84 (1994) 141-193 149

installed in a vertical section of the pipe, with 90” elbow, the flow ahead of the sampler will be
upwards flow and with the probe pointing up- strongly three-dimensional, having a helical-type
stream, the coalescence of droplets under subiso- motion, and therefore it would be difficult, if not
kinetic sampling conditions resulted in sampling impossible, to align the probe with the fluid veloc-
efficiencies significantly lower than unity. The large ity vector. Thus the sampling efficiency will be less
water droplets dropped out of the probe because than an ideal value of unity. It should be noted
the negative buoyancy forces exceeded the viscous that the helical motion will also have an effect on
drag forces that carried the droplets. Note that the distribution of droplets ahead of the sampler,
under subisokinetic conditions, one would nor- i.e. the droplets will not be uniformly distributed,
mally observe sampling efficiencies to be greater and this would obviously have an effect on the
than unity if no coalescence or gravity effect is sampling efficiency. To overcome the difficulties
present. However, when the probes were installed associated with flow-disturbing elements such as
pointing downstream, the sampling efficiency was pipe bends, etc., it is generally recommended that
significantly greater than unity under subisokinetic either the sampler should be installed at least 20
sampling conditions. This was due to the settling pipe diameters or more downstream from any
of large droplets into the probe opening. The bend, valve, or other pipe fittings, or some sort of
deviation of sampling efficiency from unity, under flow straightener should be installed [51,52]
anisokinetic conditions, was found to be signifi- upstream of the sampler.
cantly less when the probe was installed in a
horizontal section. Thus it was recommended that 2.1 S. Side-wall sampling
the probes be installed horizontally pointing Quite often, the concentration of a flowing
upstream, especially when subisokinetic sampling system is determined simply by removing the
is used. It was also observed that pulsation of flow sample through an opening in the pipe wall, as
increased the coalescing tendency of the droplets. shown in Fig. 9 [ 51,54,55]. The advantage of this
side-wall sampling is that it is a very simple and
2.1.3. Droplet bouncing efeect inexpensive technique as it uses only a small open-
According to Nasr-El-Din [ 521, particle bounc- ing in the wall and does not, in any manner,
ing effects could lead to significant sampling errors. obstruct the flow, as is the case with sampling
A bouncing effect occurs when the probe, assuming probes. The side-wall sampling, however, almost
it to be L-shaped, is thick and blunt. When the always gives erroneous results, particularly if the
droplets hit the probe wall, they lose some of their particles have a high inertia because the particles
inertia and make their way into the probe. Thus now have to change their direction by 90”. As an
the concentration of the sample may be signifi- example, Fig. 10 shows some of the experimental
cantly higher than the actual concentration, i.e. the results of Nasr-El-Din and Shook [55] for a
sampling efficiency may be greater than unity, even medium-sized sand having an average diameter of
when the sampling is carried out isokinetically. To 326 urn. The sampling efficiency is plotted as a
avoid bouncing effects, either the probe wall should function of the velocity ratio U/U, for three
be very thin or the probe should be tapered with different values of M, where M is the ratio of the
a small tip angle (e). cross-sectional areas of the side-wall opening to
the main line. It is obvious that when M is less
2.1.4. Flow structure ahead qf the sampler than unity the sampling efficiency is less than an
The structure of the flow ahead of the sampler ideal value of unity even when we have an isoki-
can have a significant influence on the sampling netic sampling situation, i.e. U/Ue= 1. Only in the
efficiency [52]; for example, if the sampling probe case of M = 1 is the sampling efficiency unity under
is installed downstream of a pipe bend, such as a isokinetic sampling.
150

(0)

Direct in-line emulsion analyzers or laboratory

emulsion analyzers could be classified either as
destructive-type or non-destructive-type analyzers.
depending upon whether or not the emulsion
undergoes any permanent degradation during the
measurement. In general, direct in-line analyzers
are non-destructive-type devices whereas labora-
tory analyzers could be destructive or non-
destructive types. One of the major disadvantages
b)
of the destructive methods is obviously the degra-
dation of the emulsion sample: therefore, if addi-

;lr:.
tional analysis of the sample is required and only
a limited amount of the sample is available, it is
preferable to use the non-destructive methods.
SIDE PORT
Table 1 summarizes the various destructive and
non-destructive techniques that are used to mea-
U’
%
sure the composition of emulsions. As indicated in
the table, some of the techniques are applicable
Fig. 9. Side-wall sampling
only to laboratory analysis of the sample. The
details of these techniques are discussed in the
1 .o
following sections.

0.8 - 3. Destructive techniques of composition

measure~lent

\%*
%=
The destructive techniques of composition meas-
urement includes titration and other chemical
methods, total organic carbon (TOC) analysis.
c)--o M=l.O centrifugation, distiliation. flash vaporization,
solvent extraction and differential scanning calo-
rimetry (DSC).
0.0 0.2 0.4 0.6 0.8 1 .o

U/U0 -

Fig. 10. Experimental results of Nasr-El-Din and Shook [SS].

Many chemical methods based on stoi~hionletric
reactions with water have been proposed and
In summary, one can obtain erroneous results if successfuliy tested for the measurement of the water
sampling is not carried out with proper under- content of emulsions [ 57-591. Examples include
standing. However, it should be noted that sam- the production of acetylene by adding calcium
pling errors in the case of emulsions are not as carbide to an emulsion sample [59], the reaction
serious as in the case of solid/liquid slurries, mainly of water with Karl-Fischer (K-F) reagent
because water and oil densities are not that much [57,60,61], and the reaction of water with acetyl
different. chloride in the presence of pyridine 157,583. Here
R. PallColloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 151

Table 1
Emulsion analyzers

Destructive type Non-destructive type

No. Name Comment No. Name Comment

Titration, and other Laboratory 1 Electrical In-line/laboratory

chemical methods
TOC analysis Laboratory 2 Microwave In-line/laboratory
Centrifugation Laboratory 3 Radiation scattering In-line/laboratory
Distillation Laboratory 4 Spectroscopic In-line!laboratory
Flash vaporization Laboratory 5 Ultrasonic In-line/laboratory
Solvent extraction Laboratory 6 Density In-line/laboratory
Differential scanning calorimetry Laboratory 7 Viscosity In-line/laboratory

we will discuss only the latter two titration reaction with water:
methods, which are more important.
&H,N.SO, + H20+CSH,NHS0,H (6)

3.1.1. Karl-Fischer method This reaction is prevented by having a large excess

The K-F method [57,60-681 is the most popu- of methanol in the system.
lar. It has been successfully used for water concen- Thus the stoichiometry of the K-F titration
trations as low as a few parts per million to involves 3 mol of pyridine, 1 mol of iodine, and
relatively very high values. 1 mol of sulphur dioxide for each mole of water.
The emulsion sample is dissolved in anhydrous The usual practice is to use both sulphur dioxide
methanol and is titrated with K-F reagent, i.e. a and pyridine in excess so that the combining
mixture of iodine, sulphur dioxide, pyridine and capacity of the reagent for water depends on its
anhydrous methanol; the iodine present in the iodine content.
reagent reacts with the water as follows: In the classical burette titration method, the
end-point is identified by the change in colour.
C,H,N*Iz + C,H,N.SO, + C,H,N + HzO+
When all water in the sample has been consumed,
2C,H5N.HI + C,H,N*SO, (4) the colour of the solution becomes yellowish owing
to the appearance of the first excess of iodine.
C,H,N*S03 + CH,OH+C,H,N(H)SO,CH,
Nowadays, there are several electrometric meth-
(5) ods available to detect the end-point [60-62,681.
In the first reaction, water is consumed to give Also, pyridine-free K-F reagents have been devel-
sulphur trioxide and hydrogen iodide; the reactants oped to overcome some of the difficulties associ-
and products exist as pyridine complexes because ated with pyridine-based reagent [ 671.
of the presence of a large amount of pyridine
(C,H,N). In the second reaction, which occurs 3.1.2. Acetyl chloride method
when a large excess of methanol is present, the Smith and Bryant [SS] successfully developed
pyridine-sulphur trioxide complex is consumed by the acetyl chloride method for the quantitative
methanol to form another complex. This reaction determination of water in organic liquids. This
is extremely important for the titration to be method is based on the fact that acetyl chloride
successful because the pyridine-sulphur trioxide reacts with water to produce 2 mol of titratable
complex can also undergo another undesirable acid in the presence of pyridine. However, acetyl
152 R. Pnl/Colloids Surfrrccs A: Phpkvchm. Eng. Aspects X4 ( 1994) 141 -I 93

chloride produces only 1 mol of titratable acid content of dilute O/W emulsions and oily
when it reacts with dry alcohol in the presence of wastewaters, with oil concentrations ranging any-
pyridine: where from a few parts per billion to 5000 ppm
[69-741. It should be emphasized that this instru-
CH,COCI - C,H,N + H,O+
ment does not measure the oil content directly; it
C,H,N - HCl + CH,COOH (7) measures the TOC content, and therefore a cali-
bration curve relating the TOC content to the
CH,COCI.C5H,N + C,H,OH-t
oil content is required. The TOC of the emulsion
C,H,N*HCl + CH,COOC,H, (8) sample is determined by measuring the total
carbon content, inorganic and organic, in one
A known amount of the emulsion sample to be
operation, and the inorganic carbon content, due
analyzed for water content is added to an excess
to dissolved carbon dioxide and carbonates etc.,
of 1.5 M reagent (a mixture of acetyl chloride and
in another. and then subtracting the inorganic
pyridine) solution in toluene and the mixture is
carbon from the total carbon. The time required
shaken vigorously. It normally takes 2 min or so
for the complete analysis of the sample is usually
for the hydrolysis reaction (Eq. (7)) to go to
completion at room temperature. The excess rea- a few minutes.
gent is then decomposed by the addition of dry A schematic diagram of a TOC analyzer is given
ethanol (Eq. (8)). The final mixture is titrated with in Fig. 11. The total carbon content is determined
standard caustic. The titration is also carried out as follows. A stream of carrier gas (oxygen or air)
with a blank system, i.e. the same quantity of is passed continuously through a combustion tube
reagent is directly decomposed by dry alcohol, containing cobalt oxide catalyst, installed in a high
without the addition of emulsion sample. The net temperature furnace (950 ‘C). The emulsion sample
increase in acidity between the sample system and is introduced into the carrier gas, and hence the
the blank system, as determined by titration with combustion tube, by injection with a microsyringe.
standard caustic, is a direct measure of the water The organic carbon is combusted catalytically to
in the emulsion. For example, if we consider N carbon dioxide. and inorganic carbon is decom-
mol of acetyl chloride complex and X mols of posed thermally to carbon dioxide. The gas stream
water then the first reaction (Eq. (7)) will produce carrying the CO, produced is passed through a
2X mol of titratable acid and the second reaction non-dispersive infrared (NDIR) analyzer, which is
(Eq. (8)) will produce (N - X) mols of acid, i.e. a sensitized specifically for carbon dioxide. Total
total of (N + X) mols of titratable acid is produced. carbon is measured from the height of a peak
For the blank run with N mols of acetyl chloride
complex, only N mols of acid are generated (Eq.
NONDISPERSIVE
(8)). The difference in the number of moles of acid INFRA-RED ANALYZER

produced is therefore X. SELECTOR VALVE

The above method has been successfully applied

to systems of widely varying water content. In
TOTAL INORGANIC
principle, one can use the method for both O/W CARBON CARBON

and W/O emulsions over a wide range of water CHANNEL CHANNEL SAMPLE
INJECTION
concentration. PORT
(HIGH TEMPERATURE (LOW TEMPERATURE
PORT

FURNACE--950-C) FURNACE*lSO’C)

3.2. Total organic carbon analyzers CARRIER

GAS
SUPPLY
(AIR OR OXYGEN)
Total organic carbon (TOC) analyzers are
widely used to determine the total organic carbon Fig. 11. Schematic diagram of TOC analyzer.
R. Pal/Colloids Surfuces A: Physicochem. Eng. Aspects 84 (1994) 141-193 153

indicated by the NDIR analyzer; the analyzer is treated with a solvent and/or demulsifier, is added
previously calibrated with use of samples of known to a centrifuge tube which is calibrated and tapered
carbon content. After having determined the total so as to allow precise measurement of emulsion
carbon, the inorganic carbon is determined by samples that have a low percentage of water.
passing the carrier gas continuously through Centrifugation is then carried out until complete
another column packed with quartz chips coated separation of oil and water phases occurs; incom-
with phosphoric acid and housed in a second low plete separation of various components can lead
temperature furnace (15O’C). The sample which to significant errors.
may have dissolved carbon dioxide or carbonates
is decomposed by the action of the acid and 3.4. Distillation
temperature, and the resulting gas is carbon diox-
ide. Owing to the absence of the catalyst and low Distillation is widely employed for the measure-
temperature, organic compounds pass practically ment of the water content of emulsions [ 63,78-851.
unaltered through the column as vapours, and Apparatus essentially of the Dean-Stark design
therefore are not detected by the NDIR analyzer. [85] is used; the emulsion sample is mixed with
There are several variations of the above tech- an appropriate organic solvent (toluene, xylene,
nique available [69]. For example, the catalyst benzene, naphtha, etc.) that is immiscible with
used in the combustion tube may not be cobalt water. Upon heating, the water vapours are codis-
oxide; some use cupric oxide (9OO’C) or platinum tilled with the organic vapours; the distillate is
(SSO’C). Also, the carbon dioxide produced may condensed in the condenser and the condensate is
be reduced to methane first by another catalyst caught in a trap (Dean-Stark trap) so designed
(nickel at 45O”C), and then the methane produced that the denser water phase collects at the bottom
can be measured with a flame ionization detector. while the lighter organic liquid flows back into the
Another technique that can be applied for the distillation flask. The water trap is calibrated so
analysis of oily wastewaters involves irradiation of that the amount of water can be determined
the sample by UV light [69], which converts the directly. The distillation procedure can be applied
organic material to carbon dioxide. The carbon equally well to both O/W and W/O emulsions. A
dioxide produced is then detected by means of the reproducibility of 0.2-0.4% is reported for emul-
NDIR analyzer. sions containing l-50% water [Sl].
In situations where emulsions also contain
3.3. Centrifugation solids, the modified Dean-Stark method [63,82]
is employed to determine the relative amounts of
Centrifugation is a commonly used laboratory water, oil and solids. Figure 12 shows a schematic
technique for the determination of the water and diagram of the modified DeanStark apparatus.
oil contents of emulsions [63,75577]. Both O/W The emulsion sample is placed in a porous thimble
and W/O emulsions, with or without solids, can that is suspended above the refluxing organic
be analyzed over a wide range of water concen- solvent. On heating, water in the emulsion sample
trations. Depending upon the viscosity and sta- is co-distilled with the solvent vapours. Thus, the
bility characteristics of the emulsion sample to be distillate, which is condensed in the condenser, is
analyzed, one may require the addition of a diluent a two-phase mixture of solvent and water. The
and/or a chemical demulsifier to the sample to water from the condensed distillate is continuously
facilitate the separation process; sometimes sulphu- separated in the graduated DeanStark trap while
ric acid is added to destroy the emulsifier before the solvent is recycled back through the thimble.
the sample is centrifuged. In the recycling process, the solvent dissolves oil
A known volume of the sample, untreated or from the emulsion sample present in the thimble
154 R. Pul/Colloids Swfaces A: Physicockrm. Eng. Aspects 84 (1994) 141LlY3

tus suitable for flash vaporization is given in

Fig. 13. The vaporization is normally carried out
at atmospheric pressure at temperatures as high as
25O”C, depending upon the composition of the
emulsion sample. The sample from the feed vessel
is pumped into the heating coil, (a glass double-
helically wound coil), where the feed is preheated
to the desired temperature: the emulsion is then
WATER TRAP flashed in a flash chamber, about 1 in in diameter
and 15 in in height. The overhead products, which
consist of a mixture of water vapour and light
hydrocarbons, are condensed and collected in the
distillate receiver, and the bottom product, the high
boiling point fraction, is collected in the bottom
product receiver. The percentage of water in the
emulsion sample is determined by measuring the
total volume of the products, distillate plus bottom
product, and the volume of water in the distillate.
Fig. 12. Modified Dean-Stark apparatus [SZ]

3.4. Solvent extraction

and carries it to the bottom of the kettle; some of
the fine solids which are not retained by the Solvent extraction is normally used to determine
thimble are also carried away by the solvent. At the oil content of dilute O/W emulsions [ 88-911
the end, the water content is read from the gradua- with oil concentrations up to 1000 ppm or so. The
ted trap; the oil/solvent mixture and fine solids method involves the extraction of oil from water
from the kettle are separated by centrifuging and into a suitable solvent (toluene, carbon tetrachlo-
the solids content is determined gravimetrically; ride, ether, chloroform, Freon, etc.). The oil content
the solids content includes both fine solids not of the solvent is then measured either with an
retained by the thimble plus solids retained by the NDIR analyzer [88,89] or with a spectrophoto-
thimble. The oil content of the emulsion sample is meter [90]. For very small amounts of oil, the
simply the difference between the total mass of the material may be absorbed by a charcoal column
original emulsion sample and the combined mass and recovered with an appropriate solvent 1911.
of water and solids.

3.5. Flush vaporization 3.7. Differential scanning calorimetry

The water and oil content of emulsions can be Differential scanning calorimetry (DSC) has
measured by separating the water from the emul- been applied recently to analyze the composition
sion using flash vaporization [ 86,871. This of emulsions [63,92,93]. In this technique, the
approach is quite effective for fairly persistent sample and inert reference material are heated or
emulsions with water concentrations as high as cooled separately at a constant rate (“C min- ‘)
80%. Both O/W and W/O emulsions can be dealt while maintaining the temperature difference
with easily. between them at zero [92-961. When any endo-
A schematic diagram of the laboratory appara- thermic or exothermic changes occur in the sample
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (I 994) 141-l 93 155

1Omm TUBE FOR

TO PRESSURE
CONTROL SYSlEY

PYROMETER WELL -u
DOUBLE HEUCU
COIL WOUND FRD
16 FEET OF &ram
TUBING
/FLASHING COILS SPACED 3Omm

CHECK SILVERED WITH

1,
FEED
VALVE llllu l/2’ WINDOWS U
CONTAINER

- PYROMETER WEL,

SILVERED
. ..^......
WATER PMSE lhaq
8 COwOLUnONS

PURE
BOILING RESIDUAL BDWUS
HE&TING MANTLE / FLUID
22mm

tc - THERMOCOUPLE PYROMETER

Fig. 13. Flash vaporization apparatus [86,87] (the figure on the right-hand side gives a detailed drawing of the flash chamber).

(for example, melting or freezing of H,O), the heat consequently, one observes an exothermic peak at
addition or removal rate to the sample also changes lower temperatures.
so as to maintain the temperature difference When the temperature of a cooled emulsion
between the sample and the reference at zero. The sample (cooled to, say, -70” C with liquid nitro-
difference in heat flow rates between the reference gen) is raised in the calorimeter, the melting of
and sample is monitored as the DSC output signal. frozen water is revealed as an endothermic peak.
When an emulsion sample is cooled in a Figure 14 shows the thermograms for a W/O
differential scanning calorimeter, a phase change emulsion containing distilled water and n-heptane/
of water is revealed as an exothermic peak in the Span 65 (sorbitan tristearate). The water droplets
DSC curve. It should be noted that the DSC are 5 pm or less in diameter [93,97]. Curve A
technique is one of the few techniques that can shows the cooling behaviour of the emulsion
distinguish “free” water, if present, from “emulsi- whereas curve B shows the heating behaviour. The
fied” water (droplets). The exothermic peak at cooling curve shows an exothermic peak at - 40” C
273 K is indicative of free water. The emulsified due to the crystallization of water in the droplets.
water undergoes subcooling, and freezes at lower The heating curve shows an endothermic peak at
temperatures depending upon the droplet size; 0°C due to the melting of ice.
156 R. Pal/Colloids Swface.s A: Physicochem. Eng. Aspcvts 84 ( 1994) 141- 193

WATER-IN-OIL
4.1 .I. Capacitunce methods
EMULSION EXO Most of the in-line techniques used for monitor-
ing the water content (water-cut) of W/O emulsions
in the crude oil pipelines are based on dielectric
constant (capacitance) measurement [98-1091.
END0
This method, however, is not generally suitable for
O/W emulsions where the conducting fluid (water)
9:
forms the continuous phase. and the capacitance
probe therefore tends to short out and provides
an erroneous indication of the water or oil
I I 1 1 1 1 J content [ 1051.
-50 -40 -30 -20 -10 0 10
The dielectric constant of W/O emulsions
TEMPERATURE (‘C)
increases rapidly with the dispersed-phase concen-
Fig. 14. DSC curves for a W/O emulsion [93,97]. tration. It also depends strongly on the excitation
frequency. The frequency-dependent behaviour of
the dielectric constant is referred to as “dielectric
relaxation” or “dielectric dispersion”. As an exam-
The area of the DSC peaks is proportional to ple, Fig. 15 shows the plot of dielectric constant
the amount of the corresponding component in for a 47% by volume Water-in-Vaseline oil emulsion
the sample. Thus one can determine the composi- at 296 K. The dielectric constant E’ is initially
tion of the emulsion sample from its DSC curve. constant up to a frequency of about 200 kHz but
it decreases rapidly with further increase in fre-
quency and it becomes constant again at high
4. Non-destructive techniques of composition
frequencies. This dielectric relaxation in W/O
measurement
emulsions, around frequencies of 100 kHz-
10 MHz, occurs because of so-called “interfacial”
The non-destructive techniques available for the
or “migration” polarization, which is sometimes
measurement of emulsion composition can be
referred to as the Maxwell-Wagner-Sillars (MWS)
grouped as follows: electrical, microwave, radiation
effect [ 1101. The presence of free charge or ions
scattering, spectroscopic, ultrasonic, and density-
and viscosity-based techniques.

4.1. Electrical methods

The electrical methods of measuring the com-

position of emulsions are very popular in the
industry. Both capacitance- and conductance-
based techniques are used widely. Because the
dielectric constant (permittivity) and electrical con-
ductivity of oil and water are very different, emul-
sions prepared from oil and water exhibit electrical
properties which are highly composition depen-
dent. Thus, by measuring the electrical properties f (kHz)-

of emulsions, one can infer the composition Fig. 15. Dielectric relaxation in a water-in-Vaseline oil emul-
accurately. sion [llO].
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-l 93 157

FREQ. f/Hz. SCALE FOR SPECIMEN B

in the dispersed water droplets is responsible for
interfacial polarization [ 110,l 1 1 1. It should be
noted that W/O emulsions generally exhibit two
dielectric relaxations; one due to interfacial polar-
ization and the other due to Debye-type dipolar
; 25
relaxation of the dispersed water phase around
lo-100 GHz. This latter effect is discussed in 20

Section 4.2, which deals with microwave measure-

ment techniques.
The relaxation frequency, which is defined as
the frequency corresponding to the half-value point
of the entire domain of interfacial polarization
relaxation, depends upon the following factors FREQUENCY, f/Hz

[ 1 lo]: dispersed-phase (water) concentration,

Fig. 17. Influence of salt concentration on the dielectric behavi-
water salinity (conductivity) and temperature. our of W/O emulsions [ 1121.
With an increase in the dispersed-phase concen-
tration 4 of W/O emulsions, the relaxation fre-
the dielectric constant E’ and loss factor E” versus
quency f0 decreases, i.e. the dielectric relaxation
frequency plots shift towards higher frequencies
phenomenon occurs at lower frequencies. The
with an increase in the salt content of the water,
relaxation frequency decreases slowly at low values
indicating an increase in the relaxation frequency.
of 4, but at high values of 4 (0.6 or above) the
The relaxation frequency also increases with an
decrease is quite rapid. Figure 16 shows a typical
increase in the temperature T.Figure 18 shows the
variation of f0 with 4. Clearly, there occurs a
plots of relaxation frequency versus l/T on a
dramatic decrease in f0 around Q = 0.6.
semilogarithmic scale for Water-in-Vaseline oil
The relaxation frequency is also known to
depend on the salinity (electrical conductivity) of
dispersed water droplets [ 1121. Figure 17 shows
7
the data of Hanai et al. [ 1121 for W/O emulsions;

80 1,
Relaxation frequency data .
for W/O emulsions

20

0
:,I$
5 L I I I

3.0 3.5 4.0

0.3 0.6 0.9

Q 103/T-

Fig. 16. Variation of relaxation frequency with dispersed-phase Fig. 18. Variation of relaxation frequency with temperature
concentration 4 for W/O emulsions [ 1121. [llO].
emulsions [llO]. The plots are linear for a given quency. The excitation frequency, however, should
value of 4. not be too low, otherwise electrode polarization
It should be noted that although the relaxation may become a problem [ 11 I 1. The selection of the
frequency is a strong function of water conductivity measuring frequency therefore requires careful con-
and temperature, the lin~iting values of the dielec- sideration of various factors: water conductivity,
tric constant of W/O emulsions, at low and high temperat~lre, water concentratjoll, and electrode
frequencies, are not significantly inAuenced by polarization. Generally, the capacitance measure-
these factors [I lo]. The limiting values, however, ments for the purpose of emulsion composition
are strong functions of water concentration. measurement are carried out at frequencies in the
Figures 19(a) and 19(b) show typical plots of range 10 kHz-15 MHz Cl 131.
the linliti~g values of the dielectric constant at low The limiting values of the dielectric constants of
and high frequencies, The data shown in Fig. 19(a) em~ilsions can be predicted from theoretical
are for low to moderately concentrated water-in- models. Several models have been developed to
vaseline oil emulsions investigated by Clausse describe the dielectric behaviour of emulsions.
[l IO]; the dispersed-phase concentration &t is Excellent reviews are given in Refs. 110 and 111.
varied from 0 to 0.525. The data shown in Here we will discuss only the Wagner [ 1143 and
Fig. 19(b) are for highly concentrated W/O emul- Hanai [ 1I 1,t 121 models, which are important.
sions [ 112j. The figures clearly denlollstrate that Wagner [ 1141 developed the following model
the l~rnit~~~dielectric constants are a strong func- for the complex permittivity E” of dilute emulsions:
tion of water concentration. Furthermore, the lim-
iting dielectric constant at low frequencies (tl) is
(9)
much more sensitive than the limiting dielectric
constant at high frequencies (E& with respect to
the water content of the emulsions (4). This is where c,* and E: are complex permittivities of the
particularly true at high values of 4. Thus it continuous and dispersed phases respectively, and
appears advantageous to carry out capacitance 4 is the dispe.t sed -p h ase concentration. Notice that
measurements for the purpose of water determina- the complex permittivity of emulsions is given as
tion at frequencies lower than the relaxation fre- E* = 6’ -j,“, where ,j is a complex number (j”=

I
Water-in-Vaseline
oil emulsions
20 r

Low
frequency

o- oT.“‘l’,.“‘,,.~“‘.i
0.0 0.2 0.4 0.6 0.5 0.6 0.7 0.8 0.9
(a) ct, (b) 4,
Fig. 19. Dependervzeof low and high frequency iin~itin~ dielectric constants on dispersed-phone concentration: (a) ~~~t~r-in-~~~sel~ne
ail em~lisions [ 1IO]; (b) W:O eln~il~ions of Wanai et al. [ 1t 2).
R. PallColloids Swfaces A: Physicochem. Eng. Aspects 84 (I 994) 141-l 93 159

- l), E’ is the permittivity (dielectric constant), and complex permittivity of emulsions:

E” is the loss factor.
The above model gives the following relation-
(16)
ships for the limiting values of the dielectric con-
stant and the electrical conductivity:
The above model gives the following relationships

1
2'5,
+ Ed-2$(& -Ed) for the limiting values of the dielectric constant

Eh= EC 2E,+ Ed+ &EC -Ed)

(10) and electrical conductivity:

K,
Er = E, -
0
KC
+
9 (Gc - e,Kd )K,$J
C2K + K, + &K - &)I2
(11)
(17)

1 (12)
1
~_-
K = K 2Kc +Kd-@(Kc -Kd)
1 K,
2K,+K,+&K,-KK,)

9 (K&c - Kc%kc&
Eh
0
Kh = Kc E, + [2e, + cd + &cc - +)I2 (13) (SI)(~)‘:=l_~
(19)

where cr and ch are the limiting permittivities at

low and high frequencies, K, and K, are limiting
electrical conductivities at low and high frequen-
cies, E, and cd are the permittivities of the con- (20)
tinuous and dispersed phases, and K, and K, are For W/O emulsions where K, >>K,, the above
the electrical conductivities of the continuous and relationships lead to the following equations:
dispersed phases.
For W/O emulsions where K, >>K,, the above $?I= 1 - (E&J - 1’3 (21)
relationships lead to the following equations:
(22)
(14)
(e&J + 2
Thus the dispersed-phase concentration of concen-
2(E, - Eh) + Ed [ 1 - @h/%)1
trated W/O emulsions can be calculated, provided
fJ= (15) that the dielectric constants for emulsion (E, or eh),
2(Ec - Ed) + Eh [ 1 - kd/‘%)I
dispersed phase and continuous phase are known.
Thus the dispersed-phase concentration 4 of W/O Again, the relationship involving the limiting
emulsions can be calculated from a knowledge of dielectric constant at low frequencies (E,) does not
the dielectric constants for the continuous phase contain the dielectric constant of the dispersed
and the dispersed phase, and the emulsion (either phase.
E, or ch). Notice that the relationship involving the Figures 20(a) and 20(b) compare experimental
limiting dielectric constant at low frequencies (E,) data with the above theoretical relationships for
does not contain the dielectric constant for the the limiting dielectric constants. The low frequency
dispersed phase. data, as well as the high frequency experimental
Wagner’s model (Eq. (9)) and all the relation- data, for W/O emulsions are in excellent agreement
ships derived from it are valid only for dilute with the Hanai relationships, Eqs. (21) and (22).
emulsions. For concentrated emulsions, Hanai The Wagner relationships (Eqs. (14) and (15))
[ 111,112] developed the following model for the underpredict the values of the dielectric constants,
160

0
0.4 0.6 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 0.6
0.0 0.2
(a) cp (a) 4,Actual 9 Actual

100
1.6
7
(Low frequency data) (High frequency doto)
-1.4
g

\ 1.2
D
e
+P-
1.0
Hanoi model Hanoi model

-I
0.4 0.5 0.6 0.7 0.8 0.9 0.6
0.4 0.5 0.6 0.7 0.6 0.4 0.5 0.6 0.7 0.6 0.3
(b) 0 QI
4, +
(b) Actual Actual

Fig. 20. Comparison of experimental da& with theoretical rela-

tions for the limiting dielectric constants of W/O emulsions: (a) Fig, 21. The ratio of predicted dispersed-phase concentration
data source, Clausse [ 1 IO]; (b) data source, Hanai et al. [ 1121. to actual concentration for W/O emulsions: (a) data source.
Ciausse [ llO]; (b) data source, Hanai et al. [ 1I?].

except at low values of 0 (d, 60.10) where the

agreement is good.
The ratio of the predicted water concentration sensors, however, are subject to coating with wax,
to actual (experimental) water concentration is dirt etc. present in the process streams. An example
plotted in Figs. 21(a) and 21 (b). Clearly, the Hanai of an invasive sensor is a pair of electrodes where
equations give accurate predictions. The water one electrode takes the form of a tubular metallic
concentration predicted from the Wagner relation- element, with a diameter less than the pipe inside
ships is much higher than the actual values. diameter, mounted coaxially inside the pipe, and
Therefore it is recommended that the Hanai equa- the other electrode takes the form of a metallic
tions be used to predict the water content of rod positioned axiall_y in the centre of the pipe by
emulsions. means of non-conductive spiders. Further examples
of invasive and non-invasive sensor designs are
4.1 .f . I. Apparatus for capncitunce measurements. given below [ 116,117]:
The apparatus for capacitance measurement con- (I) a pair of parallel metallic plates placed inside
sists of an electrode sensor and the supporting the pipe or mounted on the outside (non-
electrical circuitry. There are various designs avail- conductive) pipe wall:
able for the sensor. Both invasive (direct contact) (II) a pair of concave metallic plates placed
and non-invasive sensors are used widely, although inside the pipe or mounted on the outside (non-
Stott et al. [ 1151 recommend invasive sensors conductive) pipe wall;
because of their higher sensitivity. The invasive (IIT) a series of pairs of concave metallic plates
R. PallColloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 161

staggered spirally around the outside (non- tance signal to emulsion composition changes. In
conductive) pipe wall; many situations, the stray capacitance can be much
(IV) a pair of continuous metallic helices larger than the capacitance of the fluid being
mounted on the outside (non-conductive) pipe wall; monitored [ 1 l&119]. Thus it is important to mini-
(V) three pairs of continuous metallic helices mize the effects of strays so as to improve the
mounted on the outside (non-conductive) pipe wall sensitivity and accuracy of the measurement
and connected alternately in parallel. system. A detailed discussion of various methods
These configurations are shown schematically in to minimize the effects of stray capacitances can
Fig. 22. The continuous helix pair (IV), however, be found in Ref. 118. References 120-129 also
is preferred as it is less sensitive to flow patterns contain useful information about electrode sensors
[ 1171. Also, it should be noted that practical and the measuring circuits.
capacitance sensors always have an earthed screen In summary, the capacitance method is a feasible
to shield the sensor from external fields [ 1 IS]. and accurate method for the measurement of the
The electrical circuits for measuring the capaci- oil and water contents of emulsions. This method,
tance can be grouped into four main categories however, is generally restricted to oil external
[ 1181: a.c. bridge circuits, charge/discharge cir- emulsions (W/O emulsions).
cuits, LC and RC oscillation circuits, and reso-
nance circuits. For low frequencies below 100 kHz,
an a.c. bridge circuit is recommended; in the 4.1.2. Conductance methods
frequency range from 100 kHz to about 5 MHz, a Conductance methods [ 1303 are very similar to
charge/discharge circuit is preferred. Beyond
the capacitance methods except that the electrical
5 MHz, oscillation or resonance circuits are used.
conductivity (conductance) of the emulsion is mon-
One major problem commonly encountered with
itored instead of the dielectric constant (capaci-
capacitance measurements is the high standing
tance). This method is generally applicable to O/W
capacitance of the sensor (stray capacitance), which
emulsions only, as W/O emulsions usually have
results in low sensitivity of the measured capaci-
very low conductivity.
The electrical conductivity of O/W emulsions is
a strong function of the dispersed-phase (oil) con-
NON-CONDUCTIVE
PIPE centration. For dilute O/W emulsions, the Wagner
0) model (Eqs. (9))( 13)) leads to the following rela-
tionship for the electrical conductivity K:

K=K

c[
2(1-4J)
2+$ 1 (23)

where K, is the electrical conductivity of the con-

tinuous phase, i.e. water. It should be noted that
according to Wagner’s model, the electrical con-
ductivity of O/W emulsions is independent of the
frequency, provided that the oil is non-conducting
and non-polar (except at low frequencies where
electrode polarization may become a problem).
Fig. 22. Electrode sensors for capacitance measurement
Thus the oil concentration of O/W emulsions can
[116,117]. be calculated, provided that the electrical conduc-
162 R. Pal/Coiloids Surf&es A: Physimchcnl. Eng. Aspects 54 ( 1994) I41 -193

tivities of emulsion and aqueous phase are known: phase), Fig. 23(b) shows data for a 70% O/W
emulsion prepared from a 0.05 N KC1 aqueous
2[: 1 -w/m
(24) solution and a Nujol-carbon tetrachloride mix-
4 = -p + W/&)1 ture, and Fig, 23(c) shows data for a 70% O/W
For concentrated O!W emulsions, the electrical emulsion prepared from distilled water and nitro-
conductivity is related to the dispersed-phase (oil) benzene. As can be seen, the electrical conductivity
concentrations as follows [ 1111: of O/W emulsions is independent of the frequency
when the oil is non-polar (the Nujol-carbon tetra-
K = K,( 1 - #:’ (25) chloride mixture) but it shows a renlarkable
This equation is a special case of Hanai’s model increase at high frequencies when the oil is polar
for the complex permittivity of emulsions (Eqs. in nature (nitrobenzene).
( 17).- (20)). Thus one can predict # from knowledge Figure 24 shows a comparison of the Wagner
of K and K,: and Hanai equations discussed above (Eqs. (23)
and (25)). At low values of dispersed-phase (oil)
concentrations, there is good agreement between
It should be noted that Hanai’s model, like the two equations. At higher concentratiol~s, the
Wagner’s model, predicts the electrical conductivity Hanai equation (Eq. (25)) predicts lower values of
of O/W emulsions to be independent of the fre- emulsion conductivity compared with the Wagner
quency only if the oil is non-conductive and non- Equation, although the difference is only marginal.
polar. The conductance method of monitoring the com-
Figure 23 shows the plots of electrical conductiv- position was recently investigated by Pal [ 1301
ity versus frequency data for three different O/W for O/W emulsions flowing through a pipeline. A
emulsions [I11 t 1. Figure 23(a) shows data for a special star-shaped electrode sensor was used to
70% 0,‘W emulsion prepared from distilled water monitor the conductance of the flowing emulsions.
and a Nujol-carbon tetrachloride mixture (oil The schematic diagram of the electrode sensor
(conductance cell) is given in Fig. 25. The calibra-
tion data for the conductance cell were obtained

Oil-in-water emulsions .

- -. Wagner model
- Hanai model

IO3 IO4 lo5 lo6 10’ 0.0 0.2 0.4 0.6 0.8 1.o
FREQUENCY (cycles/set) Oil Cont.
Fig. 23. Electrical conductivity vs. frequency data for three Fig. 24. Comparison of the Wagner and Hanai models for
different O;W emulsions [ 11 I]. electrical conductivity of O:Mi emulsions.
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-l 93 163

, NON-CONDUCTIVE PIPE

Fig. 25. In-line conductance cell [ 1301 lE-1 ...‘I.“,,.‘.‘,“‘.,“”

0.0 0.2 0.4 0.6 0.8 1.0
Oil Cont.
by passing known conductivity solutions through Fig. 26. Conductance as a function of oil volume fraction for
the conductance cell. unstable (without surfactant) O/W emulsions.
Emulsions of refined mineral oil in tap water
were prepared in a mixing tank equipped with creases smoothly with an increase in oil concen-
high shear mixers. The emulsions prepared in the tration. However, at an oil concentration of about
tank were pumped through the star-shaped con- 77.53% by volume, a sudden drop in conductance
ductance cell and the conductance meter reading value occurs owing to inversion of the O/W emul-
was recorded as a function of oil concentration. sion to a W/O emulsion.
Two different sets of emulsions were prepared. The conductivity data for both sets of emulsions,
In one set, no chemical emulsifier (surfactant) was unstable and stable O/W emulsions, are compared
added so that the emulsions produced were unsta- with the Wagner and Hanai models (Eqs. (23) and
ble with respect to coalescence, i.e. the emulsions (25)) in Fig. 27. The data appear to fall closer to
readily separated into oil and water phases if left the Wagner model although the difference between
unstirred for some time. In the other set, a non- the two models is very little.
ionic emulsifier (Triton X-100) was added, and Figures 28(a) and 28(b) show the ratio of pre-
consequently the emulsions produced were very dicted to actual (experimental) values of oil concen-
stable with regard to coalescence.
For the unstable emulsions, the concentration of 1.0

oil was varied from 0 to 84.32% by volume. The

emulsions were O/W type until the inversion point 0.8
- -. Wagner model
(77.53% by volume oil) where the O/W emulsion - Hanai model

inverted to a W/O emulsion. 0.6

”
The stable emulsions were prepared by shearing
s
together known amounts of oil and aqueous surfac- 0.4
tant solution (1% by weight). The emulsions pro-
duced were the O/W type over the full range of
dispersed-phase (oil) concentration investigated
(O-72.21% by volume). No inversion of O/W emul-
sion to W/O emulsion occurred in this set. 0.0 0.2 0.4 0.6 0.8 1.0

Figure 26 shows a plot of emulsion conductance Oil Cont.

as a function of oil concentration for unstable Fig, 27. Comparison of experimental data with the Wagner and
emulsions. Initially, the conductance value de- Hanai models for the electrical conductivity of O/W emulsions.
164 R. Pul/Colloids Swfi~es A: Ph~sic~ochem. Eng. Aspects 84 ( 1994) 141-t 93

having extremely low concentrations of water. The

method is based on conductivity measurement. A
4 Wagner model schematic diagram of the sensing cell is given in
0 Hanai model
5 Fig. 29. There are two compartments separated by
*.ti a thin piece of cheesecloth, clamped as shown.
’ 1 Fluid flows through the cheesecloth from one
D
f compartment to another. The electrical conduc-
en tance of the cloth is monitored continuously while
the fluid flows through it. Electrical contact with
the cheesecloth is made at the edges and centre.
As long as the cloth is dry, its conductance remains
0.2 0.4 0.6 0.8 1 .o very low. If water is present in the feed stream, it
dJActual is removed on passage through the cloth, which
becomes more conductive. While this method is
capable of determining whether or not water is
l Wagner model present in the stream, it is unlikely that it can give
0 Hanai model accurate quantitative measurement of water
content.
Bitten [ 1321 developed a similar method for
monitoring the water content of liquid hydro-
carbon fuels. The fuel containing water is passed
through cellulose pads; the pads absorb water and
consequently their electrical conductance changes.
Using an appropriate calibration. the water content
I
of the sample is determined.
0.0 0.2 0.4 0.6 0.8 1 .o
+ Actual Lee et al. [ 1331 developed a novel in-line method
(b)
for monitoring the oil content of dilute O/W
Fig. 28. The ratio of predicted dispersed-phase concentration
emulsions (oil concentrations below 1.5%). The
to actual concentration for (a) stable O’W emulsions, and (b)
lmstable O/W emulsions. method is based on measurement of the root

tration. Both the Wagner and Hanai models predict

FLUID OUT
the oil concentration reasonably well although the t
Wagner model gives a slightly better prediction.
Thus the conductance method is a feasible and
accurate method for composition measurement of
emulsions. The method, however, is generally
applicable to O/W type emulsions only. Also, the
ELECTRODE
method is sensitive to variations in the conductivity
of the continuous phase, which is known to vary CHEESECLOT

in many applications.

4.1.3. Other conductance- and cnpncitnnce- ELECTRODE -~-

related methods t FLUID IN

Kapff [ 1311 developed a new “in-line” method

for monitoring the water content of W/O emulsions Fig. 29. Water sensing cell of Kapff [ 13 l]
R. PallColloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 165

mean square (r.m.s.) value of fluctuations in the

0
25 - OIL-IN-WATER
instantaneous conductance across the electrodes 1.0 EMULSIONS
2
placed in the flowing fluid. Figure 30 shows a
schematic diagram of the measuring principle. The
advantage of the r.m.s. measurement technique
over conventional steady state conductance meth-
ods is that it is insensitive to variations in the
conductivity of the continuous phase. The varia-
tions in the continuous-phase conductivity occur
relatively
fluctuations
very slowly whereas the conductance
across the electrodes occur very 0 5 10
J
rapidly owing to the small transit time of the CONCENTRATION OF OIL (%W/W)
discontinuous phase in the field between the
Fig. 31. Root-mean-square value of the conductivity signal as
electrodes. The slow variations in the continuous-
a function of oil concentration [ 1331.
phase conductivity are rejected using an appro-
priate filter.
uring the capacitance, and hence composition, of
Figure 31 shows a typical calibration curve for
water continuous O/W emulsions provided that
the r.m.s. method. The r.m.s. value of the conduc-
the water conductivity is not too high. The sche-
tance signal is plotted as a function of oil concen-
matic diagram of the apparatus is given in Fig. 32.
tration. There exists a linear relationship between
It consists of an invasive electrode sensor, a
r.m.s. value and oil concentration. It is important
charge/discharge transducer, and a large d.c. block-
to mention that the calibration curve is dependent
ing capacitor (1 pF) connected between the
upon the Auid velocity as well, i.e. the r.m.s. value
electrodes and the charge/discharge transducer.
of the output signal is a function of dispersed-
The purpose of the blocking capacitor is to prevent
phase concentration and fluid velocity [ 1331. This
the application of the d.c. component of the charg-
is not a serious problem because in most on-line
ing voltage pulses to the electrodes, and thus to
applications, there is also installed a flow meter to
minimize the effect of electrode polarization. The
measure the velocity (flow rate) of the emulsion.
sensor capacitance is measured by first charging it
Huang et al. [ 1221 recently developed a tech-
to a known voltage and then discharging it into
nique for monitoring the capacitance of conductive
the input of a charge measuring circuit which is at
fluids. This technique is potentially useful for meas-

INVASIVE
ELECTRODES 1 )--------
DISCONTINUOUS PHASE I
NON-CONDL!CTIVE PIPE WALL
DISTRIBUTED RANDOMLY
IN CLOUD FORMATION c ELECTRODE A
FLOW
-

0
I

PIPELINE CAN BE EITHER - ~~ELECTRODE B

ELECTRICALLY CONDUCTING
OR AN INSULATOR

CONDUCTIVITY IS MEASURED BETWEEN ELECTRODES A AND B

Fig. 30. Schematic diagram of conductivity measuring appara- Fig. 32. Schematic diagram of capacitance measuring apparatus
tus of Lee et al. [ 133). of Huang et al. [ 1221.
166 R. Pul,Kolloids SurfncesA: Pk!,.sirochrnl. Eng. Asprrts 84 ( 1994) 141-I 9.3

virtual earth potential. The relationship between

SOURCE-KLYSTRON
the measured sensor capacitance C, and the actual
fluid capacitance C, is shown to be as follows REFERENCE

[122]: LEVEL SET ATTENUATOR

SPLITTER -
C, = C,-O.StG, (27)

where t (a known quantity) is the charging pulse

duration and G, is the fluid conductance. Thus the
PHASE ANCLE-
unknown fluid capacitance can be calculated, pro- SLOTTED LINE SL

vided that the conductance of the fluid is known

through independent measurements.

4.2. Microwave measurement

Microwave techniques are receiving special

attention nowadays for the moisture measurement
of various materials [ 134-1521, including emul- ATTENUATION

sions [ 134-1431. Their greatest advantage is that

they are applicable to both O/W and W/O emul- Fig. 33. Schematic diagram of microwave measurement appara-
tus [137].
sions, and are also insensitive to variations in the
conductivity (salt content) of water. Furthermore,
the techniques are capable of distinguishing the the Hanai model covers dilute as well as concen-
type of emulsion [ 136-1381 (O/W or W/O). trated emulsions. Indeed, the Hanai model reduces
Microwaves are electromagnetic waves covering to the Wagner model in the limit 4-O.
a broad spectrum of frequencies, usually considered Recently, Thomas et al. [138] solved the Hanai
to range from 0.3 to 300 GHz. When microwaves model numerically, covering a wide range of fre-
are passed through the emulsion sample, they are quencies in the microwave region. The model (Eq.
attenuated and phase shifted. In general terms, the (16)) was rewritten as follows:
real part of the complex relative permittivity of the
( 1 - $h)(E*p = E; (E$ - c*) (28)
emulsion, E’, gives rise to a phase change whereas
the imaginary part, the term E”, gives rise to the where 4, c* and ~2 have been defined earlier, and
attenuation of a microwave beam. E,* is given as
The microwave measurement device is basically
E; = (EC~)“~/(E$
- c,*) (29)
an automatically balancing microwave bridge
which measures the phase and amplitude changes Writing the complex quantities in terms of polar
experienced by an electromagnetic wave as it passes coordinates
through the emulsion sample (see Fig. 33). From
c* = lc*\eiO (30)
the phase-shift and attenuation measurements, one
can calculate the dielectric constant E’ and the loss Ed*= [cd*[ei”d (31)
factor E” [137].
E: = IE,*l& (32)
The dielectric behaviour of emulsions over the
entire r.f. range can be described adequately by the and substituting into Eq. (2X), one obtains
Wagner and Hanai models [110-l 121 discussed
(1 - ~)1~$11~3~~;3= l~~j~~~(l~~~~~~ - iE*ie’@) (33)
earlier (Eqs. (9) and (16)). However, the Wagner
model is valid only for dilute emulsions, whereas Separating the real and imaginary parts gives the
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 167

following two relationships: (0)

(1 - $)1~*1”~ cos(8/3) + IE: IIE*Jcos(d, + 0)

- l~:)lE,*l cos(U, + dd)=O (34) - MWS

EFFECT

s 50
(1 - ~)JE*J”~ sin(0/3) + IE,*I/6*1 sin(0, + 0) k?
a \
-1~,*IIE,*Jsin(8,+8,)=0 (35)

In the above two relationships, the only two

unknowns are IE*~and 0. These equations can be
f (kHz)-
solved numerically for IE*\ and 8 at various frequen-
f

L
cies, provided that the complex permittivities of 2% (b)
the continuous phase (E,*) and dispersed phase (~2) g 25
are known. From a knowledge of (E*( and 19,one t,
2
can easily calculate the permittivity E’ and loss
factor c” for emulsions. :
J 0
Thomas et al. [ 1381 solved the above equations 1 lo* lo4 lo6 lo8
using the following relationships for the permittivi- f (kHz)--
ties of water (E;) and oil (E,*):
Fig. 34. Typical dielectric behaviour of a W/O emulsion
(adapted from Clausse [ 1 lo]).
74.325
~*=4.215+(~+~~~~)r-~
w (36)
tion. Typical dielectric behaviour of O/W emul-
E; = 2.5 (37) sions is shown in Fig. 35.
(c) The permittivity and loss factor of O/W
where f is frequency, z, is the relaxation time of emulsions are higher than those for the W/O
water (8.25 x lo-r2 s) and CLis the spreading emulsions at the same water concentration [ 1371.
parameter (0.0125). Note that the permittivity of As an example, Fig. 36 shows the permittivities
oil is assumed to be constant, independent of the and loss factors for O/W and W/O emulsions at
frequency. two different water concentrations (40% and 60%
The general dielectric behaviour of emulsions
predicted through the above equations can be
summarized as follows.
(a) The W/O emulsions exhibit two dielectric
relaxations, one due to interfacial polarization
(MWS effect) around 10 kHz-1 MHz, the other
due to Debye-type dipolar relaxation of the
dispersed-water phase around lo-100 GHz. The
typical dielectric behaviour of W/O emulsions is
shown in Fig. 34.
(b) The O/W emulsions exhibit only one dielec- I I ,I
,>
I I I I / I
lo4 IO" 10 10 10 12 10 14
tric relaxation, namely Debye-type dipolar relax-
ation of the dispersed-water phase around f(Hz)--

lo-100 GHz; the O/W emulsions do not exhibit Fig. 35. Dielectric behaviour of O/W emulsions (oil being non-
dielectric relaxation owing to interfacial polariza- conducting and non-polar).
168 R. Pd/Colloids Surfaces A: Physicochewz. Eng. Aspects 83 (1994) 141-l 93

25~~25
the difference in dielectric properties of O/W and
40x BY 20
i0 20 W/O emulsions in the microwave region can be
“OLUUE IL
WATER
15 15g exploited to determine emulsion type. They defined
2
10 w/o ‘0, a new function, referred to as dielectric modulus
lx
5 52 P, as follows:
_,-.
7
0i 0
lo8 lo9 10’“10” lo= lo”
f (Hz) - f (Hz) -
(38)

The dielectric modulus is extremely sensitive to

emulsion type and water volume fraction. It also
generally depends on the frequency.
Figure 38 shows typical plots of P versus fre-
quency. The figure clearly indicates that the dielec-
tric modulus is independent of the frequency in the
low microwave frequency range (0.1-10 GHz).
f (Hz) - f (Hz) -
Also, the dielectric modulus of O/W emulsions in
Fig. 36. Permittivities and loss factors for O/W and W/O the low microwave frequency range is much higher
emulsions in the microwave region [ 1381.
than the modulus for W/O emulsions at the same
water concentration. Furthemore, at frequencies
by volume). Clearly, the permittivities and loss
much higher than the microwave range, the modu-
factors of O/W emulsions are higher. It should
lus of both emulsion types converges to the same
also be noted that beyond a frequency of 100 GHz,
value, approximately equal to the volume fraction
the permittivities and loss factors of O/W and
of water in the emulsion.
W/O emulsions converge to the same value; this
Figure 39 shows the variation of frequency-
can be seen more clearly in Fig. 37 where permittiv-
independent dielectric modulus in the low micro-
ity is plotted as a function of water volume fraction.
wave frequency range (O.lLlO GHz) with water
Furthermore, the permittivity is almost indepen-
volume fraction for O/W and W/O emulsions. Not
dent of the water volume fraction at frequencies
only is the modulus of O/W emulsions consistently
beyond 100 GHz.
much higher than the modulus of W/O emulsions,
According to Wasan and co-workers [ 136-1381,

1 .oo 60% WATER

O/W

940, , , ( , , ( , , , , ,1
o- o.oot’
,$
’ t’ ’
,$ 10’0 10” 1o12 1o13
0.0 0.2 0.4 0.6 0.8 1 .o
Water volume fraction f(Hz) -

Fig. 37. Permittivity as a function of water concentration, at Fig. 38. Typical plots of dielectric modulus P as a function of
frequencies beyond 100 GHr (data source, Thomas et al. [138]). frequency [ 1381.
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 169

to the presence of suspended particles or droplets.

The scattering of light causes a decrease in the
intensity of the transmitted beam. The relationship
between the intensities of incident and transmitted
beams is given as follows [ 163,164]:

I = I, exp( - rl) (39)

where I, is the intensity of the incident beam, I is

the intensity of the transmitted beam after passing
through a sample of thickness 1, and r is the
0.0 0.2 0.4 0.6 0.8 1 .o
turbidity of the sample. The above relationship
VOLUME FRACTION WATER
can be rewritten as
Fig. 39. Variation of frequency-independent dielectric modulus
(frequency in the low microwave range, 0.1-10 GHz) with water A = log,,(Z,/l) = r//2.303 (40)
volume fraction for O/W and W/O emulsions [ 1381.
where A is the absorbance. Thus, by measuring
the absorbance, one can determine the turbidity of
but also the modulus is quite sensitive to water the sample. As the turbidity of the emulsion is
concentration. Thus a single measurement of the related to the concentration of the scattering par-
complex relative permittivity, and hence the dielec- ticles, one can determine the unknown concen-
tric modulus, at any frequency in the low micro- tration of the emulsion sample provided that a
wave frequency range (0.1-10 GHz) can be used calibration curve, in terms of turbidity or absor-
to determine the emulsion type and estimate water bance versus concentration, is obtained previously.
and oil contents simultaneously. It should be noted that if the sample does not
scatter light strongly, it is difficult to measure
4.3. Radiation scattering accurately the diminution in the intensity of the
transmitted beam. In such cases, it is preferred to
Radiation scattering techniques include turbidity use nephelometers [163], which measure the
measurement, neutron scattering, and measure- amount of light scattered at one or more angles to
ment of depolarization due to multiple scattering. the incident beam; the intensity of scattered light
is proportional to the concentration of scattering
4.3.1. Turbidity measurement particles. Indeed, nephelometers are now the most
Turbidimeters are often employed for the meas- commonly used devices for measuring turbidity
urement of the suspended solids concentration of [ 153,156].
dilute sludges and slurries [153-1621. They are There are two basic types of nephelometric
also used for the measurement of the oil content turbidimeter used widely for on-line measurement
of dilute O/W emulsions [ 157,161] with oil con- of turbidity [ 1531: single-beam nephelometers and
centrations ranging from 0 to 5000 mg 1-i or so. multiple-beam nephelometers. Figures 40(a) and
The water content of dilute W/O emulsions can 40(b) show schematically the principle of these
also be monitored using Turbidimeters [ 1531. As instruments. They consist of three basic compo-
turbidity measurement is a very involved topic, nents: light source, focusing lens to direct a light
here we will present only a brief discussion. For a beam from the source through the sample to be
detailed description of turbidity measurements, the monitored, and a photodetector to measure the
cited references should be consulted. transmitted or scattered light. The single-beam
If a beam of light is passed through a dispersion, nephelometers are relatively less expensive but they
such as an emulsion, the light is scattered owing suffer from the following disadvantages:
170

Fig. 40. Schematic diagram of (a) a single-beam nephelometer [ 15371, and (b) a multiple-he~tnl [ I j-i]
~~~p~~~l~~~~~~ter

(a) light source aging effect, resulting in variation signal is the active signal The turbidity of the
of source intensity (this leads to inaccurate meas- sample is determined from the following value of
urements of turbidity if the instrument is not absorbance A:
recalibrated frequently);
(b) photodetector aging effect, resulting in inac-
(41)
curate measurement of scattered light intensity;
(c) colour variation errors, caused by changes in
colour of the sample being monitored (the colour where l,,/lA, is the ratio of active to reference
acts to attenuate the light beam resulting in false signals when lamp #l is in use, and i,AJI,, is the
variations of turbidimeter readings); ratio of active to reference signals when lamp #2
(d) fouling of optical surfaces. is involved. By taking the ratio of active to refer-
Multiple beam nephelometers overcome many ence signals, most of the errors due to light source
of the problems encountered with single-beam and photodetector aging effects, colour variation,
instruments. The diagram shown in Fig. 40(b) rep- fouling etc. are minimized because both signals are
resents a four-beam in-line nephelometer. This affected identically.
instrument consists of two light sources and two There are many variations of the above two
photodetectors (A and B). The measurement of types of turbidimeter available commercially. For
turbidity involves two steps: in the first step, lamp example, in some instruments a single light source
#l is used momentarily to project a light beam is used with two or more light detectors. However,
onto the sample and photodetectors A and B the major drawback of turbidity measurements is
measure the transmitted and scattered lights; in that the scattering of light is not only dependent
the second step, lamp #2 is used momentarily to upon the concentration of the dispersed droplets
project a light beam on to the sample and, again, or particles, but it also depends upon particle size
photodetectors A and B measure the transmitted and shape, particle size distributiol~. ~avelen~th
and scattered lights. The transmitted light signal of incident light and the ratio of refractive indi-
acts as a reference signal and the scattered light ces of the particle and surrounding medium
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 171

[ 153,154,163]. Thus these variables should remain the individual particles undergoes rescattering by
constant for measurements to be accurate. other particles and this phenomenon of multiple
Furthermore, the wavelength used for the measure- scattering increases with increase in the concen-
ments should suffer minimal absorption. For exam- tration of the dispersed particles. If the incident
ple, in the case of oil and water emulsions, the light is polarized in only one plane, it becomes
optical source that emits wavelengths below depolarized owing to multiple scattering; the
0.92 urn is preferred to minimize absorption of the greater the amount of multiple scattering, the larger
emitted signal by water. is the degree of depolarization. Thus there is a
direct relationship between the degree of depolar-
4.3.2. Neutron scattering ization of incident plane-polarized light and the
In this method, a beam of “high energy” or concentration of scattering particles.
“fast” neutrons (average energy, 5 MeV) is directed Figure 41 shows a schematic diagram of the set-
into the sample [165-1671. The fast neutrons up required to measure the depolarization ratio.
interact with the sample mainly by collision with The unpolarized light from a source is polarized
hydrogen nuclei, thereby losing energy. The con- in a vertical direction by a polarizer. The plane-
centration of low energy scattered neutrons (also polarized light passes through the emulsion sample
called “slow” or “thermal” neutrons, average and a portion of the incident light is scattered by
energy about 0.025 eV) in the vicinity of the source the dispersed particles. By allowing the scattered
is a strong function of the hydrogen concentration light (depolarized) to pass through an analyzer
in the sample; the higher the hydrogen concen- prism, the horizontal and vertical components of
tration, the higher is the thermal neutron density. the light are isolated and the intensities of these
This thermal neutron density can easily be mea- components are measured by appropriate detec-
sured with an appropriate thermal neutron detec- tors. The degree of depolarization (DP) is given by
tor. If water is the only hydrogen-containing the ratio of the intensities of the vertical and
compound present in the sample, the thermal horizontal components:
neutron density is a direct measure of the water
DP = Iverticalllhorizonta, (42)
content. However, if other hydrogen is present, as
in the case of O/W emulsions, the portion of
hydrogen contained in the oil must remain con-
4.4. Spectroscopic techniques
stant for measurements to be reliable and accurate.
This neutron scattering technique has been success-
Absorption spectroscopy is widely used in the
fully tested for the measurement of the moisture
industry for the moisture measurement of various
content of butter and margarine, which are basi-
materials. It is based on the fact that when electro-
cally W/O emulsions [ 1651.
magnetic radiation is passed through a sample,
certain frequencies may be selectively absorbed,
4.3.3. Depolarization measurement
leading to attenuation of the incident beam.
Maley [168] developed a novel method for
rapid (and continuous) measurement of the concen-
TRANSMITTED
tration of particles in dispersed systems, such as SOURCE LIGHT

suspensions and emulsions. The method has been

Q____&L;:;;Q’” DEPOLARIZED LIGHT
successfully tested for dilute suspensions. It is based
upon the fact that depolarization of plane- i

polarized light occurs in the presence of multiple

scattering [ 1691.
In a dispersed system, incident light scattered by Fig. 41. Apparatus for measuring depolarization [168].
172 R. PallColloids Surfaces A: Physicochrm. Eng. Aspects 84 ( 1994) 141-l 93

The absorption of electromagnetic radiation by

some chemical species, N, of the sample can be
regarded as a two-step process [ 1701:

N + hv+N* (43)

N*+N+heat (44)

In the first step, N* represents the excited state of

the species N that results from absorption of the
photon (hv). The second step shows the relaxation
process whereby the excited species is returned to
the ground state by releasing heat (the relaxation
process may involve the decomposition of N* to
a new species). Note that the lifetime of the excited 1 2 3

state is generally short ( 1O-8-1O-9 s) and that the WAVELENGTH (microns)

amount of thermal energy released is usually very Fig. 42. Typical near-IR absorption spectrum for water [ 1761.
small.
The governing equation for the absorption of to incident light intensity I,, is plotted as a function
radiation is the Beer-Lambert law, often called of wavelength. The spectrum shows very strong
simply Beer’s law [ 1711:
absorption at 2.95 urn and progressively weaker
absorptions at 1.94, 1.43 and 1.20 urn. Most mea-
A = log,,(Z,/l) = & (45) surements, however, are made at a wavelength
of 1.94 urn.
where p is the molar absorptivity (also called the Because of the high attenuation of near-IR
extinction coefficient) and c is the concentration of energy by emulsions, it is generally not possible to
the absorbing substance (mol l- ‘). use a radiation transmission technique unless the
Absorption spectroscopic techniques are charac- water content is small and the sample is very thin,
terized according to the type of electromagnetic which is not practical for an industrial in-line
radiation employed. The techniques which are system. Therefore IR reflectance techniques are
important for the measurement of the water more commonly used.
content of emulsions are IR absorption spectro- Figure 43 shows schematically the principle of
scopy, UV absorption spectroscopy, and NMR the IR reflectance technique. Radiation from the
spectrometry. source (a quartz-halogen lamp) is focused into a
parallel beam that is projected onto the sample via
4.4.1. Infrared absorption spectroscopy a rotating wheel equipped with IR filters. The
IR absorption techniques are quite popular in purpose of the filters is to provide the means of
the industry for the moisture measurement of selecting narrow bands of IR light at wavelengths
materials [63,172-1761, including emulsions corresponding to the “reference” and “absorption”
[ 1721. These techniques utilize the fact that water wavelengths. The reference IR radiation is such
molecules selectively absorb IR radiation between that it undergoes negligible absorption by water;
the wavelengths of 1 and 3 urn. The absorption is for example, a wavelength such as 1.7 um is less
accompanied by vibrational excitation of the -OH sensitive to water absorption. The pulses of light
groups in water. Figure 42 shows a typical near IR arriving on the sample are both partially reflected
absorption spectrum for water. The transmittance, and partially absorbed. A portion of the reflected
defined as the ratio of transmitted light intensity I light is collected by a mirror onto an IR detector.
R. PallColloids Swfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 173

4.4.2. Ultraviolet absorption

The UV absorption technique can be used to
determine the total organic carbon (TOC) content,
PRIMARY
DETECTOR
and hence the oil content, of oily wastewaters
(dilute O/W emulsions) [ 177,178]. When a beam
of UV radiation (about 254 nm) is passed through
a sample, attenuation of the transmitted beam
LAMPJ FILTERi
occurs owing to the absorption of UV radiation
WHEEL
by organic carbon. The attenuation of UV radia-
tion is proportional to the TOC content of the
sample. As an example, Fig. 44(a) shows a plot of
UV absorbance at 254 nm as a function of the
TOC content of sewage samples. The plot is linear
Fig. 43. Principle of IR reflectance technique [ 1751
in accordance with the Beer law.
A UV instrument for measuring the TOC
content of oily wastewaters is available commer-
cially [ 1771. A schematic diagram of the working
The logarithmic ratio of the signals of the absorp- principle of the commercial instrument is shown
tion wavelength and reference wavelength is a in Fig. 44(b). It consists of a low pressure mercury
direct measure of the water content. Note that lamp as the light source. The light beam, after
since the reference and absorption beams have passing through the emulsion sample, is received
taken the same optical path, factors other than the by a dichroic mirror fixed at an angle of 45” to
variation of water content, which can affect the the incident beam. The mirror splits the beam into
signal levels, will have equal influence on both two parts: the UV light (254 nm) which falls on a
beams, and thus will cancel out when the ratio is UV detector, and the visible light (about 510 nm)
calculated. which is deflected by a 90” angle, as shown, and
The main drawback of IR backscattering tech- falls on a visible detector. The logarithmic ratio of
niques is that they tend to measure the water the output signal from the UV detector to the
content near the surface of the sample because the output signal from the visible light detector is
IR beam can only penetrate a short distance. Thus determined electronically; this ratio is directly pro-
the emulsion sample must be homogeneous for portional to the TOC content of the sample. It
measurement to be accurate. should be noted that the visible light signal acts
A method to overcome the above difficulty asso- as a reference signal, and therefore errors due to
ciated with backscattering techniques is to use an fouling of optical surfaces, aging of light source
attenuated total reflectance (ATR) technique and detectors, etc. are minimized because both
[ 172,174]. In this method, an IR beam having two signals, reference and measuring, are affected in
wavelengths, measuring and reference, is made to the same manner.
pass through a crystal (with a number of total
internal reflections) with which the sample is in 4.4.3. Nuclear magnetic resonance
direct physical contact. For each reflection at the The NMR technique is often used for the meas-
crystal surface, part of the measuring wavelength urement of the water content (actually hydrogen
energy is absorbed by water in the sample. Thus content) of many materials [63,179-1811, includ-
the logarithmic ratio of the attenuated IR signal ing emulsions [ 631. It is a spectroscopic technique
to reference signal is a direct measure of the water which utilizes the fact that hydrogen nuclei can
content in the sample. have two different energy levels when placed in a
 R. Pd/Colloids Swfuces A: Physicochern. Eng. Aspects 84 ( 1994) 141-193

o.or
0
I

50
I

100 150
I I

200
60 ORGANIC CARBON (mg/L)

Fig. 44. (a) Ultraviolet absorbance (at 254 nm) as a function of the TOC content of sewage samples [ 1771. (b) Schematic diagram of
the UV absorbance technique [ 1771

strong homogeneous magnetic field. Under these

conditions, the nuclei can change their energy level
from a lower to a higher state by absorption of
radiation at characteristic frequencies (Larmor fre-
quency). This absorption of energy is a direct
measure of the total hydrogen content of the
sample.
Figure 45 shows the basic arrangement of the
NMR instrument. A homogeneous unidirectional
magnetic field is produced by a permanent magnet.
The sample is located in the centre of the magnetic Fig. 45. Schematic diagram of the NMR apparatus [ 1791.

field, as shown in the figure. Because of the presence

of the external magnetic field, the hydrogen nuclei by [171]:
in the sample can have two nuclear spin states and
vr = (4.258 x 107)H (46)
the Larmor frequency (rr) at which a transition
between these two states can take place is given where H is the magnetic flux density in SI units
R. PallColloids Surfaces A: Phpsicochem. Eng. Aspects 84 (1994) 141-193 175

(tesla). The sample is also subjected to an alternat- system, such as an emulsion or a suspension.
ing magnetic field by an r.f. oscillator at some fixed Consequently, there occurs a reduction in the
frequency; 60 MHz is the most commonly used intensity of the transmitted signal. This attenuation
frequency. The magnetic flux density of the external of the ultrasonic signal is related to the concen-
unidirectional field is then varied with the help of tration of the scattering particles [ 182,183]. Thus
a low frequency sweep generator. Because the by measuring the attenuation of the ultrasonic
Larmor frequency depends on the magnetic flux signal one can infer the concentration of the dis-
density (see Eq. (46)), a variation in magnetic flux persed phase.
density results in variation of the Larmor fre- Recently, Balachandran and Beck [ 1831 applied
quency. When the magnetic flux density is such the sound-scattering technique to solid/liquid
that the Larmor frequency matches the frequency slurry flows in pipes. They found that the attenua-
of the r.f. oscillator, nuclear resonance occurs and tion of the transmitted ultrasonic signal varies with
energy is absorbed. The energy absorbed by the three parameters: mean fluid-velocity u, mean par-
protons from the r.f. oscillator (when resonance ticle size ;IIP, and dispersed-phase concentration.
occurs) produces modulation of the amplitude of Figure 46 shows some of their results on a log-log
the r.f. signal; this modulated signal is measured plot. The meter reading, given in millivolts, is a
to give the hydrogen content of the sample.
measure of the depth of modulation of the ultra-
The main drawback of the NMR technique is
sonic signal. Figure 46(a) indicates that for a given
that it measures the total hydrogen content (rather
particle size and a given mean flow velocity, the
than water content) of the sample; this includes
attenuation of the ultrasonic signal increases with
the hydrogen content of the oil. If the composition
an increase in the dispersed-phase concentration.
of the oil changes, the NMR signal would change
The attenuation also increases with the mean fluid
even when the water content remains the same.
One method of circumventing this difficulty is to
use pulsed NMR [ 1811 rather than continuous-
wave NMR. The pulsed NMR technique utilizes
the difference in relaxation times of the hydrogen
nuclei in water and in oil, and is therefore selective
to the hydrogen content in water.

4.5. Ultrasonic techniques

Several ultrasound-based techniques have been DISPERSED-PHASE CONCENTRATION

proposed, and successfully tested, for measuring (X%4.)

the concentration of the dispersed phase in emul-
sions and suspensions [ 182-2081. Both O/W and
W/O emulsions can be monitored accurately. The
ultrasonic techniques for concentration measure-
ment can be grouped into three different categories:
V=l.9 m/s
sound-scattering-based techniques, sound-velocity- . 1.2% by wt.
based techniques and sound-vibration-potential-
based techniques. 1000
10 100
MEAN PARTICLE DIAMETER
4.5.1. Sound-scattering-based techniques (v)-

Ultrasonic waves are scattered by dispersed par- Fig. 46. Attenuation of transmitted ultrasonic signal for
ticles as the waves travel through a dispersion solid/liquid slurry flows [ 1831.
velocity for a given particle size and solids concen- measure the percentages of oil and water even if
tration. When the mean fluid velocity and solids the emulsion is not homogeneous; (b) it works well
concentration are held constant, the level of attenu- for both O/W and W/O emulsions: (c) it is non-
ation increases with the mean particle size, as intrusive; (d) it is relatively inexpensive.
shown in Fig. 46(b). Since the plots shown in The acoustic velocitimeter technique for measur-
Fig. 46 are linear (on a logPlog scale). one can ing the composition of emulsions utilizes the fact
write the following relationship to describe the that the sonic velocity is different in diflerent
attenuation of the ultrasonic signal: materials, e.g. oil and water, and therefore the
sonic velocity in emulsions is a function of emulsion
R, = A(G)"(C)~@,) (47)
composition (percentages of oil and water).
where R, is the meter reading, C is the dispersed- In pure single-phase fluids, the sonic velocity I’,,
phase concentration (wt. %) and A, a, h and e are is given by
constants. Therefore in a constant particle size
v; = (c’p/c’& (48)
system, if the mean fluid velocity remains constant,
the attenuation of the ultrasonic signal will depend where (?p/Zp),s is the rate of change of pressure p
only upon the dispersed-phase concentration. with density p at constant entropy s. By definition.
A variation of the above method involves the the adiabatic compressibility p is given as
measurement of scattered sound, instead of the
transmitted sound. The intensity of scattered sound (49)
is directly related to the concentration and particle
size of scattering species [ 184,203]. The effect of Combining the above two equations gives
particle size, however, can be decoupled using
multiangle ultrasonic scattering [203]. Lenn [ 1841 V” = (pp) ‘I2 (50)
recently applied a multiangle ultrasonic-scattering
In the case of emulsions, the above equation is
technique to solid/liquid dispersions. It was found
modified as follows [ 200,202]:
that although the intensity of sound measured at
a scattering angle of 170” is a function of particle T/* = (p$,)) ls2 (51)
size and concentration, the ratio of 1O- to 170 ’
where I/* is the velocity of sound in the emulsion,
scatter is a function of particle size alone. Thus it
pe is the effective density and be is the effective
is possible to measure the particle concentration
adiabatic compressibility of the emulsion. The
as well as the particle size using multiangle sound-
effective properties can be taken as follows
scattering techniques. However, it should be noted
[ 200.2021:
that little or no experimental data are available on
the application of such techniques to oil and water Pe = dv)d + ( 1 - 4)P, (52)
emulsions.
Be = 4Pd + ( 1 - 4vc (53)

4.5.2. Sound-velocity,-based techniques where pc and p, are the density and compressibility
Acoustic velocitimeters have been successfully of the continuous phase, y, and Pd are the density
used for the measurement of the oil and water and compressibility of the dispersed phase, and 6
contents of emulsions [ 182,188,192-196,203]. is the dispersed-phase concentration. Upon substi-
They are the most popular among the ultrasonic tuting Eqs. (52) and (53) into Eq. (51). the
techniques for the composition measurement of following result is obtained:
emulsions, and they have been studied extensively
(V,/V*)’ = ( 1 + B$)( 1 + 04) (54)
by several workers [1922204]. The advantages
claimed for this technique are as follows: (a) it can where V, is the ultrasonic velocity in the
R. PullColloids Swfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 177

continuous phase, B = (pd - p,)/p,, and D= Model 2

(Pd - ~c)/~c. Using Eq. (50), the factor D can be
expressed as

(55) Model 3

where V, is the ultrasonic velocity in the dispersed v*=#v,+(l -4)v, (59)

phase. Consequently, Eq. (54) can be rewritten as
Models 2 and 3, which are simple mixing rules,
(v,IV*J2 = t- 1 - ti( 1 - Pd/Pc)l give better prediction than model 1 (Urick equa-
tion). It is interesting to observe that all three
x c1 - & 1 - PcT/E/Pdv;)l (56)
models are independent of emulsion type, that is,
This is a well-known result, originally derived by at a given concentration of xylene, the sonic veloc-
Urick [200]. According to the above equation, ity predicted for a xylene-in-water emulsion is the
V* = V, when pc = pd and V, = V,. same as that predicted for a water-in-xylene emul-
Figure 47 shows the variation of sonic velocity sion. It should further be noted that the above
in xylene and water emulsions as a function of models could be rewritten in terms of travel times
xylene volume fraction. The sonic velocity of ultrasonic waves through the emulsion and
decreases with an increase in xylene volume frac- pure phases.
tion because xylene has a lower sonic velocity than
water. The figure also shows the following three Model 1
models for the sonic velocities in emulsions:
Model 1
1
-=‘ll/(l-Pd/i’i)l[l-~(l-~)]
(V”)Z v: -4(1 -;)(I -;$)(1 -$)I:’
(60)
(57)
?(l -:)(I -E$)

‘;; 1500Tr----- Xylene and water _

Model

t* - t
2

’
E
141
l Experimental data ($2 (61)
-- model 1 td - h
.... model 2
- model 3 Model 3

4= ($2) (62)

where t* is the travel time of an ultrasonic wave

in emulsion, and t, and td are travel times of
ultrasonic wave in the continuous phase and the
dispersed phase. Thus one can predict the dis-
0.0 0.2 0.4 0.6 0.8 1.o persed-phase concentration 4 from a knowledge of
Xylene volume fraction t*, t, and rd.
Fig. 47. Variation of sonic velocity in emulsions of xylene and
Tavlarides and co-workers [ 193-1961 recently
water (data source, Bonnet and Tavlarides [ 1921). proposed another model for the prediction of the
178 R. PalICol1oid.s Surfaces A: Physicochem. En<?.Aspwts 84 ( 1994) 141-l 93

dispersed-phase concentration in emulsions. The

. Xylene and water
model takes into account refraction and reflection
phenomena of sound at the droplet oil/water inter- 0.8

face. The model is given as follows:

Model 4

t* - t,
(I= (63)
gdtd - gctc

where g, and g, are factors defined as

g~=l+sc1-(l-~~)‘l?l-~
0.2 0.4 0.6 0.8

(a) True Xylene Vol. Froc.

X [ 1 _ (1 _ ;,2)5/21 _ $2 for y<l (64)

or

g,= 1 +++(I -kji!2]+$2

x [I-(1-$)r;2]+(l-$);;2for 731 (65)

gd = ( I/$)‘)[ 1 - ( 1 - 1’2)3’2] for ‘id 1 (66)

or
B
nL: 0.4 A
g, = l/,$2 for 731 (67)
0.0 0.2 0.4 0.6 0.8 1 .o
where (b) True Xylene Vol. Frac.

ultrasonic velocity in the dispersed phase Fig. 48. (a) The volume fraction of xylene predicted from model
:‘= 4 vs. the true xylene volume fraction. (b) The ratio of the
ultrasonic velocity in the continuous phase
predicted volume fraction of xylene (from various models) to
(68) the true xylene volume fraction.

Model 4 is claimed to be superior to other models,

some cases. For example, Fig. 49 gives a plot of
although in some cases it can give large errors.
predicted versus true volume fraction of the organic
For example, Fig. 48(a) shows a plot of the volume
phase for two different emulsion systems, namely
fraction of xylene predicted from this model versus
toluene-in-water emulsions and 30 vol.% tributyl
the true xylene volume fraction. Clearly, the model
phosphate (TBP)/n-dodecane in 0.2 M nitric acid
underpredicts the volume fraction of xylene. The
solution emulsions. The values predicted from
differences in the prediction capacities of the four
model 4 are in excellent agreement with experimen-
models can be seen more clearly in Fig. 48(b),
tal values.
where the ratio of predicted to true volume fraction
is plotted for xylene and water emulsions. Model 4.5.3. Ultrasound-llibration-potelztinl-bnsr~
3 gives the best prediction and Model 4 gives the techniques
least accurate prediction of xylene volume fraction. In the ultrasound vibration potential (UVP)
Model 4, however, gives excellent predictions in technique. a sound wave is introduced into the
R. PallColloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 179

0.5. . . . . , . . . . , . * . . ,
.l Toluene -in-water emulsions

:
.n (30% TBP/n-dodecane)-in-
(0.2 M nitric acid solution)
2 0.4 -
a
.-0
6 r

WATER-IN-OIL

0.0 0.1 0.2 0.3 0.4 0.5

True Vol. Frac. of Organic phase

Fig. 49. The predicted volume fraction of the organic phase

from model 4 vs. the true volume fraction of the organic phase
(data source, Tsouris and Tavlarides [ 1941).
0 2 4 6 8 10 12

(b) VOLUME OF EMULSIFIED WATER (%)

dispersed system (emulsion or suspension) by a
piezocrystal transducer [ 20552083 (see Fig. 50(a)). Fig. 50. (a) Schematic diagram of the ultrasound vibration
potential (UVP) apparatus [205]. (b) Variation of the UVP
The dispersed particles follow the oscillatory
signal with the water content of W/O emulsion [107].
motion of the liquid, caused by sound waves, only
if the density of the particles is exactly the same as
scopic electric field, which is measured as the UVP
the density of the continuous liquid phase. In
signal. The UVP signal is known to be a function
emulsions, the densities of the continuous and
of the dispersed-phase concentration d. The theo-
dispersed phases are usually different, and conse-
retical relationship relating the UVP signal with
quently there occurs a relative motion between the
various parameters is given as follows [2052081:
particles and the continuous-phase liquid. Given
that the particles are charged (more often than not, UVP = j*$ (69)
they are) and that the ionic atmosphere surround-
i = (&/p&*/K*) (70)
ing the particles carries a net charge equal in
magnitude but opposite in sign to that of the where Ap is the density difference between the
particles, the negative and positive charges experi- dispersed and continuous phases, K* is the com-
ence different velocities relative to the electrodes. plex conductivity of the emulsion, and p* is the
(It should be noted that the ionic atmosphere frequency-dependent electrophoretic mobility.
surrounding the particles is part of the continuous- Figure 50(b) shows a typical calibration curve
phase liquid.) The resulting separation of the oppo- for W/O emulsions. As expected, the UVP signal
site charges produces an electric field. Thus, the is quite sensitive to water concentration. Thus it is
introduction of the sound wave to the emulsion possible to measure the water content of W/O
effectively produces oscillating dipoles and the net emulsions by measuring the ultrasonic vibration
result of the oscillating electric fields is a macro- potential. The same is true for O/W emulsions.
However, the main drawback of this technique is (usually gamma-rays or X-rays) passing through
that the UVP signal is not only dependent on 4, the emulsion.
but it also depends on several other factors as From the Beer-Lambert law it follows that
indicated in Eqs. (69) and (70) and some of
dl Jdx = - j11 (73)
these factors are known to vary in the actual
applications. where I is the intensity of radiation, .Y is the
distance travelled by the beam through the mate-
4.4. Density-bused techniques rial, and n is the absorption coefficient. Integration
of Eq. (73) yields [226]:
The oil and water contents of emulsions can be
In (III,) = - /1,.x, - jl,.~, - pd.yd (74)
determined from the measurement of the density
of emulsion [209-2151. The density of the emul- where jlW, jlc and jld are the absorption coefficients
sions (p,) can be expressed as of the wall material. continuous phase and dis-
persed phase respectively, I, is the intensity of the
Pe = C&l + ( 1 - C&w (71) unattenuated source beam, I is the intensity of the
where p0 and pw are the densities of oil and water, transmitted (attenuated) beam and .x is the distance
respectively, and C, is the oil volume fraction. The penetrated by the beam in the various materials
above equation can be rewritten as involved. Equation (74) can be rewritten as

ln(I/I’) = - jt,x, - jldsd (75)

C, = (P, - PW)l(P, - PW) (72)
where I’ is the intensity of the transmitted (attenu-
Thus one can determine C,, and consequently C,,
ated) beam observed with an empty pipe, i.e. when
from a knowledge of the emulsion density.
only wall material is in the path of the beam. From
The density of emulsions can be measured using
Eq. (75) and the relationships
either off-line or on-line methods. The off-line
methods require a discrete sample of the emulsion 4, = .Q/(S, + .Yd) (76)
to be drawn from the pipe or other process equip-
ment; the density is then measured in the labora- x, + .yd = D (77)
tory using a pycnometer or any other suitable where D is the inside diameter of the pipe and 4
technique. Obviously the accuracy of off-line meth- is the dispersed-phase concentration, it follows that
ods depends to a large extent on representative
sampling. ln(l/Z’) = - [jl& + jj,( 1 - 4)lD (78)
The on-line measurement of density is quite
As the adsorption coefficients of materials are
popular in the industry [ 209-2321. The important
approximately proportional to their densities. Eq.
on-line techniques are as follows:
(78) can be modified to
(a) radiation gauge [223-2321;
(b) vibrating densitometer [209-211,2155221]; ln(Z/I’)= - a[p,& +p,( 1 -4)]D= -ap,D (79)
(c) continuous weighing [218];
where a is a constant and p, is the emulsion
(d) pressure-drop measurement [212-214,222,
density. Thus the attenuation of the beam, mea-
2261.
sured as ln(l/Z’). is proportional to the density of
the emulsion. Furthermore. it can easily be shown
46.1. Rudiation gauge than Eq. (79) leads to the following relationships:
A radiation gauge measures the density of emul-
sions in pipe flow by measuring the attenuation of
(80)
the strength of a concentrated beam of photons
R. PaliColloids Swfaces A: Physicochern. Eng. Aspects 84 (1994) 141-l 93 181

Some of these vibrating elements are shown sche-

matically in Fig. 52.
The natural frequency of the vibrating element
where I, and I, are the attenuated intensities when is dependent upon the total mass of the oscillating
only the continuous phase or dispersed phase alone system which includes the vibrating element and a
is present in the pipe (sample cell). Therefore one fixed volume of the fluid within or surrounding
can calculate the emulsion density or the dispersed- the element. A change in the density of the fluid
phase concentration 4 by first measuring the inten- changes the mass of the oscillating system and
sities of the attenuated beam with the pipe full of hence the natural frequency of oscillation. If m is
oil and water respectively, and then by measuring the mass of the vibrating element, V is the volume
the intensity of the attenuated beam with the pipe of fluid taking part in the oscillation, and k is the
full of emulsion. system stiffness, then the natural frequency is given
The major components of the radiation density by
gauge are (a) a source of radiation (y-rays or
X-rays), and (b) a detector of transmitted radiation.
(82)
Two main sources of y-rays are caesium-137, which
has an energy of 662 keV, a half-life of 30 years,
where fis the natural frequency of oscillation, and
and a linear absorption coefficient in water of
pe is the emulsion density. The period of oscillation,
0.086 cm-’ at room temperature, and thulium-170,
T, is defined as
with an energy of 84 keV, a half-life of 127 days,
and a linear absorption coefficient of 0.18 cm- 1 in
water at room temperature. Instruments are avail-
able commercially from several companies. It
should be noted that many newly developed y-ray Taking the square of this expression and inserting
(and X-ray) techniques utilize multiple beams,
47r2v 4rc’m
spreading radially from a single source. Figures A’=- and (84)
51(a) and 51 (b) show schematic diagrams for some k “=T
of the multiple-beam densitometers [231]. The use one obtains
of multiple-beam techniques minimizes the errors
due to inhomogeneities in the emulsion sample. T2 = A’p, + C’ (85)

or

4.6.2. Vibrating densitometers ,o,=A+CT’ (86)

The basic component of vibrating densitometers
is a vibrating element which is made to vibrate at where A= - (C’/A’) and C = (l/A’) are calibration
its natural frequency [209-211,21SP221] by an constants. A more general form of the above
electromagnetic drive. The vibrating elements of equation [216,220] is given as
different shape and design are available in the
p,=A+BT+CT’ (87)
market; they can be broadly classified into two
groups: The calibration constants (A, B and C) are deter-
(a) immersion-type vibrating elements, such as mined using fluids of precisely known density.
a thin rectangular metallic vane, a thin-walled The vibrating densitometers could be installed
metallic cylinder, and a tuning fork; either in the by-pass loop or in the main line. They
(b) flow-through vibrating elements, such as a are generally very accurate and are much less
vibrating tube. expensive than the radiation density gauge.
182 R. PallCalloids Surfuces A: Physicochem. Eng. Aspects 84 (1994) 141-193

PHOTO~ULTIPLIER TUBE -__

IMPEDENCE- ,- COLLIMATOR ASSEMBLY (LVID)

MATCHING STA

VERTICAL ARRAY OF
SEVEN DETECTOR
UNITS LEAD-DISK COLLIMATOR
(X-RAY EXIT THROUGH CENTRALLY
POSITIONED VERTICAL SLIT)

CABLES TO TAPE-
RECORDERCONTR
UNIT POWER
SUPPLiES

VERNIER SCALE

OUTLINE A REPRESENTS RIGID DETECTOR SUPPORT PLATE

OUTLINE El REPRESENTS LEAD SHIELOING BOX (CUT AWAY FOR BEAM ENTRY)

/h”
TOP OF PIP-
LEAD COLLIMATOR

DETECTOR PHOTOMULTIPLiER -
TUBE (TYP. 3 PLACES) \

i CENTER OF ACTIVE
PORTION OF SOURCE

---’ MOUNTING CLAMP

DETECTOR CASKS
(TYP. 3 PLACES)

(b)

Fig. 51. (a) Multiple-beam X-ray densitometer 12311. (b) Three-beam gamma densitometer 12311

change is measured by a weight detector. From

Figure 53 illustrates the principle of a con- the weight measurement and the known volume
tinuous-weighing densitometer [218]. The device (fixed), the density of the fluid is calculated.
consists of a U-tube, which is pivoted on flexible It should be noted that straight-tube designs,
end couplings. The fluid whose density is to be instead of the U-tube, are also available in the
measured flows through this U-tube; if the density market.
of the fluid changes, the total weight of the U-tube While continuous-weighing densitometers are
assembly (U-tube and fluid inside) changes. This quite accurate, their major drawback is that they
R. PdColloids SurfucesA: Phpsicochem. Eng. Aspects 84 (1994) 141-193 183

tical pipe (or inclined pipe) or in a vertical U-tube

[212-214,222,226]. For a straight vertical pipe of
length L and internal diameter D, the pressure
drop is given by

AP = +J+Y(3~]

where AP is the difference between upstream and

(b) downstream pressures,f is the friction factor, 7 is
the average fluid velocity and g is the acceleration
due to gravity. In many applications, the frictional
loss is negligible. Therefore the pressure drop is
given by

AP = (&)A (89)

Thus the density of the emulsion can be calculated

as follows:

me= ApI@ (90)

The U-tube arrangement is preferred when the

frictional-loss component is not negligible. In the
upward limb
Fig. 52. Schematic diagrams of various vibrating elements used
for density measurement.
AP, = pe[gL+ 2j-($‘] (91)

whereas in the downward limb

AP, = p.[ -gL+2j&‘]

From Eqs. (91) and (92)

API - AP, = (2gL)p, (93)

or

VP, - AP2)
Pe = (94)
2gL

Fig. 53. Principle of a continuous-weighing densitometer [218]. Thus the density of the emulsion can be calculated
from the pressure-drop measurements in the two
are bulky and cumbersome and they must be limbs of a vertical U-tube.
installed horizontally. In summary, the density-based techniques are
feasible and accurate techniques for measuring the
4.6.4. Pressure-drop measurement composition of O/W or W/O emulsions. However,
The density of flowing fluids can be inferred they function only as long as the densities of oil
from pressure-drop measurement in a straight ver- and water are different. If the densities of oil and
184

8 . . ..>....!...‘I....
W/O emulsion
data of Pal

6 -

-0.0 0.2 0.4 0.6 0.8 0.0 0.1 0.2 0,3 0.4
(a) Oil Cont. (b) Water Cow.

Fig. 54. (a) Viscosity as a function of oil volume fraction for Newtonian O/W emulsions [233]. (b) Viscosity as a function of water
volume fraction for Newtonian W/O emulsions 12331.

water are the same, emulsion density becomes power-law model:

independent of the composition.
7 = K~J” (95)

4.7. Visccrsity-based techniques where z is the shear stress, : is the shear rate, and
K and II are power-law constants. Newtonian
Emulsions generally behave as Newtonian Auids behaviour is a special case of the power-law behavi-
at low-to-moderate concentrations of dispersed our in that II= 1 and K = ~7(the viscosity of the
phase in that the viscosity is constant, independent fluid). The power-law index 17is generally less than
of shear rate or shear stress. The viscosity of unity in the case of non-Newtonian emulsions.
Newtonian emulsions is directly related to Figures 55(a) and 55(b) show the variation of K
the dispersed-phase concentration. For instance, and R with dispersed-phase concentration $ for
Fig. 54(a) shows the plot of viscosity as a function O/W 12341 and WiO [235] emulsions respec-
of oil volume fraction for Newtonian O/W emul- tively. The power-law parameter K increases with
sions [233]. The viscosity increases with an 4 in a regular manner and, thus, could be used to
increase in the dispersed-phase volume fraction. monitor the composition of the emulsion.
Figure 54(b) shows a similar plot for Newtonian However, it should be pointed out that neither
W/O emulsions [233]. Again, the viscosity K nor the viscosity of the emulsions is a unique
increases with the dispersed-phase concentratio~l functi~)n of the dispersed-phase concentratioll.
in a well-defined manner. Thus it is possible to They depend upon the viscosities of the oil and
calculate the volume fraction of dispersed phase water, droplet size distribution. nature and concen-
(and continuous phase) from a knowledge of the tration of surfactant, etc. For instance, Fig. 56
viscosity of the emulsion. shows the experimental data of Van der Waarden
However, at high dispersed-phase concentra- (I2361 and Eilers [2?7 J for O/W emulsions. The
tions, emulsions usually exhibit non-Newtonian data are plotted as relative viscosity. defined as the
pseudoplastic behaviour and the apparent viscosity ratio of the viscosity of the emulsion to that of
decreases with an increase in the shear rate. the continuous phase. versus dispersed-phase (oil)
The shear stress/shear rate behaviour of non- concentration. Even though the continuous phase
Newtonian emulsions is often described by a (water) in the two systems has a similar viscosity,
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 185

NEWTONIAN
OIL-IN-WATER
EMULSIONS
data of Pal
1000 r

0.0 0.3 0.6 0.9 0.0 0.3 0.6 0.9

(a) 4, 9
VAN DER WMRDEI

0.0 0.2 0.4 0.6 0.8 ;'.o

a-
Fig. 56. Relative viscosity qr as a function of the dispersed-
phase concentration 4 for O/W emulsions; comparison of the
data of van der Waarden [236] and Eilers [237].

The pressure-gradient-based viscometers involve

0.6
a pressure-drop measurement over a straight sec-
tion of a pipe. For laminar flow of a Newtonian
fluid, the well-known Hagen-Poiseuille law gives
0.0 0.3 0.6 0.9 0.0 0.3 0.6 0.9
(b) 6 @ 7cD”(AP/L)
Y= (96)
Fig. 55. Variation of power-law parameters K and n for (a) 128Q
O/W emulsions 12341, and (b) for W/O emulsions [235].
where AP is the pressure drop over a straight
the viscosities of the emulsions are widely different. length Lof a pipe of internal diameter D, 11is the
It is therefore important to obtain experimentally emulsion viscosity, and Q is the volumetric flow
a calibration curve (K versus 4) for the emulsion rate through the pipe. For non-Newtonian power-
system whose composition is to be monitored. law fluids, such as concentrated emulsions, the
There are a variety of in-line viscometers avail- pressure drops are measured in two successive
able [238-2521 which can be used to measure K pipes of different diameters (see Fig. 57). The
or the viscosity of an emulsion on a continuous power-law parameters are then determined from
basis. These include pressure-gradient viscometers the following relationship [253]:
[ 244,245], coaxial cylinder viscometers [ 244,245],
vibrating-element viscometers [ 240-2441, rota- (97)
tional viscometers of the helical impeller type
[ 2461, gyrating-element viscometers [ 247,248], where ~~ ( = APD/4LJ is the wall shear stress, and
rotating-disc viscometers [ 25 11, and falling-piston v is the average velocity in a given pipe. The slope
(or ball) viscometers [ 2521. However, pressure- of log r, versus log( 8/D) gives II and the intercept
gradient-based viscometers, coaxial cylinder vis- leads to K.
cometers and vibrating-element viscometers are Coaxial cylinder viscometers are also used
more popular in the industry [244,245]. widely in the industry. Most commercial versions
186

of such viscometers cover a wide range of speeds vibrating element is ll~easlired and this is a simple
(hence shear rates). Consequently, they are quite function of the square root of the product of
suitable for both Newtonian and non-Newtonian viscosity and density [340]:
emulsions. They can be installed in the main ,,--
pipeline or in the by-pass loop of the main line. Power = tr(\~~),r~,) (98)
The vibrating-elelnent viscometers are also pop-
where (I is a calibration constanl. and /‘e and J/~
ular in the industry. The heart of these viscometers
are the emulsion density and viscosity respectively.
is a vibrating sphere or a cylindrical rod, which is
The instrument electronics are designed to display
immersed in the flowing fluid (see Fig. 58). The
the value of the product (~,r/,_,).If the density of the
element is made to vibrate at its natural frequency
emulsion is known, its viscosity can be calculated.
by an electromagnetic device that is hermetically
The vibrating viscometer could be installed
sealed from the fluid. The vibrations are usually of
either in the by-pass loop or in the main line. The
small amp~it~ide, about I pm or so. The power
main limitation of the vibrating viscometer is that
required to maintain constant the amplitude of the
it measures the product of density and viscosity,
- DRIVER COIL rather than the viscosity.
DETECTOR COIL
/
\ I

5. Summary and conclusions

There are a variety of destructive and non-

TRANSDUCER - ‘- CROSSARM
BLOCK destructive techniques available for the mcasure-

,/
--SHEATH ment of the oil and tvater content of erntl~siolls.
Among the destructive techniques of emulsion
composition measurement, the K-F titration

&
I/ I,

“4 method. centrifugation and distillation arc quite

popular in the industry. particularly in the petro-
SENSOR TIP leum industry. The K-F method is the fastest
among all these techniqutx. It requires every small
Fig. 58. Schematic diagram of a \libr~itill~-spher~ \ iswmcter satnple volumes and is very accurate. Further-
[230]. more. little or no sample treatment is required.
R. Pul,Colloids Swfirces A: Physicwhem. Eng. A.sprct.s 84 (I YY4) l41-lY3 187

Distillation and centrifugation methods are also or density based. The capacitance-based methods
accurate but they require much larger sample are quite accurate but are generally restricted to
volumes and they are not as rapid as the K-F oil-external (W/O) emulsions. Also, the capacitance
method. Also, these methods require sample treat- probes, particularIy the invasive type, are subject
ment (dilution with solvent and/or addition of to coating with wax, dirt etc. present in the process
demulsi~er). For the c~ntrifu~~tion method to pro- streams. The density-based techniques are apphca-
duce accurate results. it is illlportant to ensure that ble to both 0,/W and W/O emulsions. However, it
complete separation of oil and water is achieved should be noted that although the density of the
and that there is no unresolved emulsion left at emulsion can be measured quite accurately, the
the interface between the separated oil and water density-based technique can be used to calculate
phases. In the case of difficult to break emulsions, the oil and water content of emulsions only as
flash vaporization may be used. Like distillation, long as a significant difference exists in the densities
this method is not so fast and it requires larger of oil and water. The technique cannot be used at
sample volumes. Solvent extraction and TOC ana- all if oil and water have the same densities.
lyzers are generally restricted to very dilute O/W The conductance-based techniques are generally
emulsions (oily wastewaters). DSC is the only applicable to water-external (O/W water) emul-
technique capable of distinguishing free water from sions as oil-external (W/O) emulsions have very
emulsified water (droplets) in W/O emulsions. One low conductivity. They can be very accurate pro-
can determine the amounts of free water (if present) vided that the conductivity changes are solely
and emulsified water in emulsions from the area due to emulsion composition. In many practical
of the DSC peaks. However, few experimental applications, however, conductivity changes in
results are reported on the application of this emulsions occur due to the variation of water
technique to emulsion composition measurement. conductivity (salinity) so that the effect of composi-
All destructive techniques have the disadvantage tion is masked. Therefore conductance probes are
that the emulsion sample is destroyed during the rarely used to monitor the oil and water content
process of measurement. Thus, if a limited sample of fluids produced from oil wells.
is available and one requires additional analysis of Among the radiation scattering techniques, tur-
the sample, destructive techniques are not recom- bidity measurements are popular. However, they
mended. The other major drawback of the destruc- are restricted to very dilute O/W and W/O emul-
tive techniques is that they are generally off-line sions. The major drawback of turbidity measure-
(laboratory) techniques and they require extraction ments is that the turbidity depends on several
of a sample from the pipeline or other process factors, other than the composition, such as droplet
equipment. The accuracy of the destructive meth- size, droplet size distribution, and refractive index
ods of emulsion analysis, therefore, depends largely of individual components. These factors should
on how representative the sample is. remain constant for turbidity measurements to pro-
The non-destructive techniques of emulsion duce accurate results. Other radiation-scattering-
composition measurement, however, are generally based techniques (Neutron scattering and depolar-
capable of both on-line and off-line analysis and ization measurement) have not been studied in any
they can be used for cotltil~~lous lnonitoring of detail. Future work should focus on experimental
process streams. Among these techlliques, capaci- evaluation of these techniques.
tance-based techniques and density-based tech- Spectroscopic techniques are potentially very
niques are quite popular in the industry. Most net useful but little experimental data are available to
oil computers at present employed in the oil indus- determine the full range of the applicability and
try for the measurement of the oil and water the limitations of these techniques.
production of oil wells are either capacitance based Among the ultrasonic techniques for composi-
18X R. Pa//Co/loids Surfaces A: Physicochem. Eng. Asprcts 84 ( 1994) 141-f 93

tion measurement, the techniques based on mea- been a great success in continuous production
surement of sonic velocity (acoustic velocitimeters) streams despite variations in salinity, temperature
in emulsions are the most versatile. They are and emulsion state (well-dispersed or not so well-
applicable to both O/W and W/O emulsions. dispersed). The instrument is capable of measuring
However, it should be noted that although the the full range of water concentration (O-100%) and
acoustic velocitimeters can accurately measure the oil concentration (O-100%). The accuracy is
velocity of ultrasonic waves in emulsions, they can reported to be within 20.05%.
be used to determine the oil and water content of Finally, it is recommended that future research
emulsions only if a significant difference exists in work should focus on the experimental evaluation
the velocities of sound in the pure oil and water of various techniques available so as to determine
phases. If the sonic velocities in the oil and water their full range of applicability and limitations.
phases are the same, one cannot use this technique There is still a general lack of experimental data
for emulsion composition measurement. available on these techniques. Experimental work
The UVP-based technique for composition dealing with comparison of the various techniqes
measurement is of limited value as the UVP signal would be most useful.
is not a unique function of emulsion composition;
the UVP signal depends on several other factors, Acknowledgement
in addition to the percentages of oil and water in
the emulsions. Sound scattering techniques have Financial support from the Natural Sciences
not been studied in any detail; little experimental and Engineering Research Council (NSERC) of
data are available to determine the accuracy and Canada is gratefully acknowledged.
limitations of these techniques for emulsion
applications.
References
The viscosity-based techniques for the measure-
ment of emulsion composition are of limited value
mainly because the viscosity is not a unique func- 1 R. Raghavan and S.S. Marsden. SPE J., 11 (1971) 153.
2 F. Steinhauff. Petroleum. 25 (1962) 294.
tion of the composition. The viscosity of emulsions 3 K.J. Lissant, in D.O. Shah and P.S. Schechter (Eds.),
depends on several other factors, such as droplet AIChE Symposium on Improved Oil Recovery by
size and distribution, nature and concentration of Surfactant and Polymer Flooding, AIChE, New York,
1976, pp. 933100.
surfactants, presence of electrolytes, etc. Thus these
4 F.W. Jenkins. in L.C. Case (Ed.), Water Problems in Oil
factors should remain constant for meaningful Production ~ An Operator’s Manual, 2nd edn..
measurements of emulsion composition. Petroleum Publishing Co., Tulsa, OK. 1977, Chapter 13.
5 J.L. Lummus, US Patent 2 661 334. 1953.
The microwave-based technique is relatively new
6 B.B. Williams, J.L. Gidley and R.S. Schechter, Acidizing
but is becoming increasingly important in indus- Fundamentals, SPE Henry L. Doherty Series. Vol. 6,
trial applications. This technique has been studied SPE-AIME, Dallas. TX, 1979, pp. 29-37, 58 -59.
extensively by Wasan and co-workers [136-l 381. 7 R. Simon and W.G. Poynter, J. Pet. Technol.. 20 (1968)
1349.
The major advantages of the microwave technique 8 C.D. McAuliffe, R. Simon and C.E. Johnson, US Patent
are as follows: (a) it is applicable to both O/W and 3 380 531, 1968.
W/O emulsions; (b) it is capable of distinguishing 9 CD. McAuliffe, Fullerton. R. Simon, Whittier and C.E.
Johnson, US Patent 3 467 195, 1969.
the type of emulsion, i.e. O/W or W/O; (c) it is
10 A.H. Beyer and D.E. Osborn, Down-hole emulsification
insensitive to variations in the salinity/conductivity for improving paraffinic crude production, Paper SPE
of the emulsions. Recently, Texaco Inc. [ 1341 has 2676, presented at the 44th Annu. Fall Meet. SPE and
AIME, Denver. CO, 1968.
developed a commercial version of the microwave
11 S.S. Marsden and R. Raghavan, J. Inst. Pet., 59 (1973) 273.
technique for the purpose of measuring oil and 12 F.M. Mourits, L. Kurucz and R.J. Scoular, Heavy oil
water production of oil wells. The instrument has emulsion research at Saskatchewan Research Council,
R. PallCol1oid.s Sur@ces A: Phpsicochern. Eng. Aspects X4 (1994) 141-l 93 189

Paper no. 14, presented at the Annu. Tech. Meet. South Encyclopedia of Emulsion Technology, Vol. 3, Marcel
Saskatchewan Section, Pet. Sot. CIM, Regina, Sask., 1987. Dekker, New York, 1988, Chapter 7.
13 J.L. Zakin, R. Pinaire and M.E. Borgmeyer, J. Fluids 38 G. Allyn, in K.J. Lissant (Ed.), Emulsions and Emulsion
Eng., 101 (1979) 100. Technology, Part 1, Marcel Dekker, New York, 1974,
14 E.V. Seymour, US Patent 3 520 313, 1970. Chapter 7.
15 R. Simon and W.G. Poynter, US Patent 3 519 006, 1970. 39 L. Benali, J. Colloid Interface Sci., 156 (1993) 454.
16 R. Simon, CD. McAuliffe, W. Poynter and H. Jennings, 40 C. Gimson, Meas. Control, 22 (1989) 79.
US Patent 3 487 844, 1970. 41 G. Agar, P. Clewis and C. Spencer, Hydrocarbon Process.,
17 W.C. Simpson and H.J. Sommer, US Patent 2 981 633. 72 (1993) 55.
1961. 42 C. Angelidou, E. Keshavarz, M.J. Richardson and
18 D.P. Rimmer, A.A. Gregoli, J.A. Hamshar and E. Yildirim, G.J. Jameson, Ind. Eng. Chem. Process Des. Dev., 16
Adv. Chem. Ser., 231 (1992) 2955312. (1977) 436.
19 R.M. Decker and D.L. Flock, J. Can. Pet. Technol., 27 43 A.H. El Sherif, Hydrocarbon Process., 72 (1993) 171.
(1988) 69. 44 J. Bitten, Anal. Chem., 40 (1968) 960.
20 D.A. Alvarado and S.S. Marsden, SPE J., 19 (1979) 369. 45 C. Tsouris and L.L. Tavlarides, Ind. Eng. Chem. Res., 32
21 G.G. Binder, N.J. Clark and CD. Russell, US Patent 3 (1993) 998.
203 517 (1965). 46 B.A. Lawton and R. Pethig, Meas. Sci. Technol., 4 (1993)
22 S. Torrey, Emulsions and Emulsifier Applications - 38.
Recent Developments, Noyes Data Corporation, NJ,
1984. Sunzpling
23 S. Friberg, Food Emulsions, Marcel Dekker, New York, 47 R.A. Furness, Meas. Control, 20 (1987) 7.
1976. 48 R. Northedge, in Proc. 1st Int. Conf. Multiphase Flow,
24 K.J. L&ant, Emulsions and Emulsion Technology, Stavanger, Norway, October 1987, Paper 18.
Marcel Dekker, New York, 1974. 49 N.W. King, G.L. Purfit and W.C. Pursley, in Proc. 2nd
25 H. Bennett, J.L. Bishop and M.F. Wulfinghoff, Practical Tnt. Conf. Multiphase Flow, London, UK, June 1985,
Emulsions ~ Applications, Vol. 2, Chemical Publishing Paper 53.
Co., New York, 1968. 50 R.C. Gold and S.G. Simmons, J. Pet. Technol., 36
26 P. Becher, Emulsions: Theory and Practice, Krieger, (1984) 2002.
Malabar, FL, 1977, Chapter 8. 51 J.H. Rushton and J.G. Hillestad, Sampling of non-
27 M.J. Lynch and W.C. Griffin, in K.J. Lissant (Ed.), homogeneous flow in pipes, in API Proc., American
Emulsions and Emulsion Technology, Part 1, Marcel Petroleum Institute, Washington, DC, 1964, pp. 5177534.
Dekker, New York, 1974, Chapter 5. 52 H.A. Nasr-El-Din, in P.N. Cheremisinoff (Ed.),
28 SE. Friberg and I. Kayoli, in M. El-Nokaly and Encyclopedia of Environmental Control Technology ~
D. Cornell (Eds.), Microemulsions and Emulsions in Wastewater Treatment Technology, Vol. 3, Gulf
Foods, ACS Symp. Ser. 448, American Chemical Society, Publishing Co., Houston, TX, 1989, Chapter 14.
Washington, DC, 1991, Chapter 2. 53 H.A. Nasr-El-Din, C.A. Shook and M.N. Esmail, Can.
29 J. Schut, in S. Friberg (Ed.), Food Emulsions, Marcel J. Chem. Eng., 62 (1984) 179.
Dekker, New York, 1976, Chapter 8. 54 H.A. Nasr-El-Din, C.A. Shook and M.N. Esmail, Can.
30 E. Graf and H. Bauer, in S. Friberg (Ed.), Food Emulsions, J. Chem. Eng., 63 (1985) 746.
Marcel Dekker, New York, 1976, Chapter 7. 55 H.A. Nasr-El-Din and C.A. Shook, Int. J. Multiphase
31 I.S. Shepherd and R.W. Yoell, in S. Friberg (Ed.), Food Flow, 12 (1986) 427.
Emulsions, Marcel Dekker, New York, 1976, Chapter 5. 56 K. Iinoya, Kagaku Kogaku Ronbunshu, 34 (1970) 69.
32 N.J. Krog, T.H. Riisom and K. Larsson, in P. Becher
(Ed.), Encyclopedia of Emulsion Technology, Vol. 2, Titration and other chemical methods
Marcel Dekker, New York, 1985, Chapter 5. 57 J. Mitchell and D.M. Smith, Aquametry: Part I ~ A
33 B.A. Mulley, in K.J. Lissant (Ed.), Emulsions and Treatise on Methods for the Determination of Water, 2nd
Emulsion Technology, Part 1, Marcel Dekker, New York, edn., Wiley, New York, 1977.
1974, Chapter 6. 58 D.M. Smith and W.M.D. Bryant, J. Sot. Chem. Ind., 57
34 S.S. Davis, J. Hadgraft and K.J. Palin, in P. Becher (Ed.), (1935) 841.
Encyclopedia of Emulsion Technology, Vol. 2, Marcel 59 R.W. Roberts and A. Fraser, J. Sot. Chem. Ind., 29
Dekker, New York, 1985, Chapter 3. (1910) 197.
35 C. Fox, in K.J. Lissant (Ed.), Emulsions and Emulsion 60 J. Mitchell and D.M. Smith, Aquametry: Part III (The
Technology, Part II, Marcel Dekker, New York, 1974, Karl Fischer Reagent) -- A Treatise of Methods of the
Chapter 13. Determination of Water, 2nd edn., Wiley, New York, 1977.
36 R.E. Ford, in A.L. Smith (Ed.), Theory and Practice of 61 R.C. Moore, Meas. Control, 19 (1986) 33.
Emulsion Technology, Academic Press, London, 1976, 62 Y.Y. Liang, Anal. Chem., 62 (1990) 2504.
Chapter 13. 63 R.J. Mikula, Adv. Chem. Ser., 231 (1992) 799129.
37 H.A. Bampfield and J. Cooper, in P. Becher (Ed.), 64 D.A. Skoog and D.W. West, Fundamentals of Analytical
190 R. Pal/Colloids Surfaces A: Physicochetn. Eng. Aspects 84 ( lYY4) 141 193

Chemistry, 2nd edn., Holt, Rinehart and Winston, New 358/82, Institute of Petroleum. London. 1982; ASTM
York, 1969, Chapter 32, pp. 750-752. D4006-81, American Society for Testing and Materials.
65 Syncrude Analytical Methods for Oil Sand and Bitumen Philadelphia, PA, 1981: API 10.2. American Petroleum
Processing, Syncrude Canada Ltd., Edmonton, Alta., Institute, Washington. DC, 1987.
1979. 85 E.W. Dean and D.D. Stark, lnd. Eng. Chem.. 12 (1920)
66 Water in Crude Oil by Karl Fischer Volumetric Titration; 486.
IP 356/87, Institute of Petroleum, London, 1987; ASTM
4377-87, American Society for Testing and Materials, Flash vaporization
Philadelphia, PA, 1987. 86 T.H. Gouw, Anal. Chem.. 49 (1977) 1887.
67 I. Nordin-Andersson and A. Cedergren, Anal. Chem., 57 87 J.A. Lockwood, R.L. LeTourneau. R. Matteson and
(1985) 2571. F. Sipos, Anal. Chem., 23 ( 1951) 1398.
68 AS. Meyer and C.M. Boyd, Anal. Chem., 31 (1959) 215.
Solvent extraction
88 Horiba Instruments Inc.. In Tech, 32 (1985) 13.
89 T. Takahashi, T. Miyahara and Y. Nishiraki, J. Chem.
69 R. Briggs, J.W. Schofield and P.A. Gorton, Water Pollut.
Eng. Jpn., 12 ( 1979) 394.
Control, 75 (1976) 47.
90 W.T. Strickland, SPE J.. 20 (1980) 175.
70 C.E. Van Hall, J. Safranko and V.A. Stenger, Anal. Chem.,
91 K.J. Lissant, Demulsification Industrial Applications.
35 (1963) 315.
Marcel Dekker, New York. 1983. p. 42.
71 C.E. Van Hall and V.A. Stenger. Anal. Chem.. 39 ( 1967)
503.
Differential scnnning calorimrtr~~
72 R.H. Jones and A.F. Dogeforde, ISA Trans., 7 (1968) 267.
92 D. Senatra, G.G.T. Guarini. G. Gabriclli and M. Zoppi,
73 N.J.M. Van Ham. L.A. Behie and W.Y. Svrcek, Can.
J. Phys. (Paris), 45 (1984) 1159.
J. Chem. Eng., 61 (1983) 541.
93 J.L. Cavallo, D.L. Chang and F.Z. Saleeh, AIChE Symp.
74 N.D. Sylvester and J.J. Byeseda, SPE J., December
Ser. 277, 86 (1990) 16.
(1980) 579.
94 G. Widmann and R. R&en. Thermal Analysis. Htithig,
Heidelberg, 1987, pp. 31-33.
95 J.W. Robinson. Instrument Analysis. Marcel Dekker. New
75 Water and Sediment in Crude oil by the Centrifuge York, 1982, p. 432.
Method; IP 359/82, Institute of Petroleum, London, 1982; 96 F.W. Fifield and D. Kealey. Principles and Practice of
ASTM D4007-81, American Society for Testing and Analytical Chemistry, 3rd edn., Blackie. London. 1990.
Materials, Philadelphia, PA, 1981; API 10.3. American pp. 455-458.
Petroleum Institute, Washington, DC, 1987. 97 P.J. Bruggeller, Dispersion Sci. Technol.. 3 ( 19X2) 395.
76 Syncrude Analytical Methods for Oil Sand and Bitumen
Processing, Syncrude Canada Ltd., Edmonton, Alta., Electrical methods
1979. 98 G.H. Ennis and W.W. Ennis. US Patent 2 617 299. 1946.
77 K.J. Lissant, Demulsification - Industrial Applications, 99 W.J. Warren, Cerrito. R.L. W&on. Oakland and F.C.
Marcel Dekker, New York, 1983, p, 42.
Schneider, US Patent 3 006 189, 1961.
100 J.B. Rosso, US Patent 3 385 108, 1966.
Distillation 101 J.B. Cox and J.R. StoltL, US Patent 3 693 435. 1972.
78 J.A. Lockwood, R.L. LeTourneau, R. Matteson and 102 E.W. Gass, J. Hammond. P. Lorenzino and J.E. Thomas.
F. Sipos, Anal. Chem., 23 (1951) 1398. US Patent 3 748 446. 1973.
79 T.H. Gouw, Anal. Chem., 49 (1977) 1887. 103 C.W. Zimmerman and G.V. Copland, US Patent 3 X02
80 D.A. Skoog and D.M. West, Fundamentals of Analytical 261, 1974.
Chemistry. 2nd edn., Holt, Rinehart and Winston, New 104 C.W. Zimmerman and G.V. Copland. US Patent 3 839
York. 1969, pp. 747-749. 911, 1974.
81 J. Mitchell and D.M. Smith, Aquametry: Part I ~ A 105 G. Agar, P. Clewis and C. Spencer, Hydrocarbon Process.,
Treatise on Methods for the Determination of Water, 2nd 72 (1993) 55.
edn., Wiley, New York, 1977, p. 277. 106 W.J. Warren, J. Pet. Technol., 14 ( 1962) 1207.
82 Syncrude Analytical Methods for Oil Sand and Bitumen 107 R.S. Wood, Oil Gas J.. December (1958) 102.
Processing, Syncrude Canada Ltd., Edmonton, Alta., 108 R.A. Furness, Meas. Control. 20 ( 1987) 7.
1979. 109 T.D. Callinan, R.M. Roe and J.B. Romans. Anal. Chem..
83 Water in Petroleum Products and Bituminous Materials 28 (1956) 1911.
by Distillation; IP 74/82, Institute of Petroleum, London, 110 M. Clausse, in P. Becher (Ed.). Encyclopedia of Emulsion
1982; ASTM D95-83, American Society for Testing and Technology, Vol. 1. Marcel Dekker. New York, 1983.
Materials, Philadelphia, PA, 1983; API 10-2, American Chapter 9.
Petroleum Institute, Washington, DC, 1987. 111 T. Hanai, in P. Sherman (Ed.). Emulsion Science.
84 Determination of Water in Crude Oil by Distillation; IP Academic Press, London, 1968. Chapter 5.
R. PallColloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-I 93 191

112 T. Hanai, T. Imakita and N. Koizumi, Colloid Polym. 145 E.R. Gates and A.J. Taylor, Meas. Control, 4 (1971) 312.
Sci., 260 (1982) 1029. 146 K. Lindberg and U. Ternstrom, Meas. Control, 3 (1970)
113 R.J. Colvile, Meas. Control, 20 (1987) 101. T33.
114 K.W. Wagner, Arch. Elektrotech. (Berlin), 2 (1914) 371. 147 M. Kent and E. Kress-Rogers, Trans. Inst. Meas. Control,
115 A.L. Stott, R.G. Green and K. Seraji, J. Phys. E, 18 8 (1986) 161.
(1985) 587. 148 K. Carr-Brion, Trans. Inst. Meas. Control, 9 (1987) 3.
116 M.S.A. Abouelwafa and E.J.M. Kendall, IEEE Trans. 149 J.G. Anderson, Meas. Control, 22 (1989) 82.
Instrum. Meas., 29 (1980) 24. 150 F. Thompson, Meas. Control, 22 (1989) 210.
117 G.A. Gregory and L. Mattar, J. Can. Pet. Technol., 151 T.F. Jones, J. Haynes and T. Jones, Meas. Control, 24
12 (1973) 48. (1992) 78.
118 SM. Huang, A.L. Stott, R.G. Green and MS. Beck, 152 R.Y. Lee, In Tech, 31 (1984) 9.
J. Phys. E, 21 (1988) 242.
119 D. Alien, Processing, August-September (1989) 55. Radiation scattering
120 C.G. Xie, A.L. Stott, A. Plaskowski and MS. Beck, Meas. 153 D. Lex, Instrum. Control Sys., 66 (1993) 41.
Sci. Technol., 1 (1990) 65. 154 R.J. Simms, ISA Trans., 11 (1972) 146.
121 S.M. Huang, J. Fielden, R.G. Green and MS. Beck, 155 R. Briggs and K.T.V. Grattan, ISA Trans., 3 1 ( 1992) 111.
J. Phys. E, 21 (1988) 251. 156 G.A. Lee, Meas. Control, 15 (1982) 450.
122 S.M. Huang, R.G. Green, A.B. Plaskowski and M.S. Beck, 157 P. Extance and G.D. Pitt, Meas. Control, 17 ( 1984) 343.
J. Phys. E, 21 (1988) 539. 158 R. Briggs, ISA Trans., 21 (1982) 59.
123 J.J.M. Geraets and J.C. Borst, Int. J. Multiphase Flow, R.D. Vanous, ISA Trans., 20 (1981) 91.
159
14 (1988) 305. 160 T.R.P. Gibbs, Optical Methods of Chemical Analysis,
124 M.-H. Chun and C.-K. Sung, Int. J. Multiphase Flow, 12
McGraw-Hill, New York, 1942, Chapter 3.
(1986) 627.
161 V.A. Price, In Tech, 38 (1991) 33.
125 E.A. Hammer and R.G. Green, J. Phys. E, 16 (1983) 438.
162 R.S. Farinato and R.L. Rowell, in P. Becher (Ed.),
126 T.M. Shi, S.J.R. Simons, F.J. Dickin and R.A. Williams,
Encyclopedia of Emulsion Technology, Vol. 1, Marcel
Colloids Surfaces, 77 (1993) 9.
Dekker, New York, 1983, Chapter 8.
127 D.M. Smith and J. Mitchell, Aquametry: Part II ~ A
163 F.P. Hochgesang, in I.M. Kolthoff and P.J. Elving (Ed%),
Treatise on Methods for the Determination of Water, 2nd
Treatise on Analytical Chemistry, Part I: Theory and
edn., Wiley, New York, 1977.
Practice, Vol. 5, Interscience, New York, 1964, Chapter 63.
128 D.A. Doughty, Anal. Chem., 49 (1977) 690.
164 G.W. Castellan, Physical Chemistry, 3rd edn.,
129 C. Gimson, Meas. Control, 22 (1989) 79.
Benjamin/Cummings, Menlo Park, CA, 1983,
130 R. Pal, Ph.D. Thesis, University of Waterloo, Ont., 1987.
pp. 9288935.
131 S.F. Kapff, Oil Gas J., 77 (1979) 133.
165 G.D. McPherson, Instrum. Rev., 12 (1965) 464.
132 J. Bitten, Anal. Chem., 40 (1968) 960.
166 H.A. Slight, Meas. Control, 22 (1989) 42.
133 K.T. Lee, M.S. Beck and K.J. McKeown, Meas. Control,
167 R.Y. Lee, In Tech, 31 (1984) 9.
7 (1974) 341.
168 L.E. Maley, Meas. Control, 5 (1972) 3.
169 E. Oster, Chem. Rev., 43 (1948) 357.
Microwoe measurements
134 Jiskott Autocontrol (Kent, UK), Meas. Control, 25 (1992)
Spectroscooic techninues
253; Processing, April (1993) 18.
170 D.A. Skoog and D.M. West, Fundamentals of Analytical
135 D.A. Doughty, Anal. Chem., 49 (1977) 690.
Chemistry, 3rd edn., Holt, Rinehart and Winston, New
136 J.P. Perl, H.E. Bussey and D.T. Wasan, J. Colloid Interface
York, 1976, Chapter 22.
Sci., 108 (1985) 528.
171 G.W. Castellan, Physical Chemistry, 3rd edn.,
137 J.P. Perl, C. Thomas and D.T. Wasan, J. Colloid Interface
Sci., 137 (1990) 425. Benjamin/Cummings, Menlo Park, CA, 1983, Chapter 24.
138 C. Thomas, J.P. Per1 and D.T. Wasan, J. Colloid Interface 172 R.J. Colvile, Meas. Control, 20 (1987) 101.
Sci., 139 (1990) 1. 173 D.B. Scrivens, D.W. Kent and R.A.E. Sargeant, Meas.
139 J.P. LePetit, G. Delbos, A.M. Bottreau, Y. Dutuit, Control, 3 (1970) T23.
C. Marzat and R. Cabanas, J. Microwave Power, 12 174 H.A. Slight, Meas. Control, 22 (1989) 42.
(1977) 335. 175 LB. Benson, Meas. Control, 22 (1989) 45.
140 J. Mitchell and D.M. Smith, Aquametry: Part III ~ A 176 D.C. Brunton, Meas. Control, 3 (1970) T89.
Treatise on Methods for the Determination of Water, 177 R. Briggs, J.W. Schofield and P.A. Gorton, Water Pollut.
Wiley, New York, 1977, Chapter 1, pp. 23-24. Control, 75 (1976) 47.
141 R.J. Colvile, Meas. Control, 20 (1987) 101. 178 R. Briggs and K.T.V. Grattan, ISA Trans., 31 ( 1992) 111.
142 S.K. Aggarwal and R.H. Johnston, IEEE Trans. Instrum. 179 L. Young, Meas. Control, 22 (1989) 54.
Meas., 35(4) (1986) 630. 180 H.A. Slight, Meas. Control, 22 (1989) 42.
143 R.W. Parsons, SPE J., 15 (1975) 302. 181 P.N. Gambhir and A.K. Agarwala, J. Am. Oil Chem.
144 H.S. Slight, Meas. Control, 22 (1989) 42. Sot., 62 (1985) 103.
192 R. PaliColloids Swfuces A: Physicochem. Eng. Aspects X4 ( 3994) 141 ~~193

Ultrasonic techniques 217 T. Cousins. Meas. Control. 25 (1992/93) 292.

182 R.C. Asher, Meas. Control, 15 (1982) 169. 218 P. Mason, Meas. Control, 25 (1992.!93) 297.
183 W. Balachandran and MS. Beck, Trans. Inst. Meas. 219 J. Gray, Meas. Control, 25 (1992/93) 303.
Control, 2 (1980) 181. 220 B.W. Balls. ISA Trans.. 20 ( 1981) 67.
184 C.P. Lenn, in P.J. Lloyd (Ed.), Particle Size Analysis 221 R.C. Gold and S.G. Simmons, J. Pet. Technol.. 36
1985, Wiley, Chichester, 1987, p. 647. (1984) 2002.
185 J.A. McKnight and A. Clare, Meas. Control, 23 (1990) 222 R. Van der Veen, in Proc. Hydrotransport 2, BHRA. UK.
208. 1972, Paper J2.
186 M.P. Wiltshire, Meas. Control, 22 (1989) 51. 223 D.R. Carlson, Meas. Control. IO (1977) 83.
187 A.K. Holmes and R.E. Challis, Colloids Surfaces, 77 224 A.J. Livingston, M&C Meas. Control. 24 (1990) 111.
(1993) 65. 225 N.N. Kondic and G.D. Lassahn. ISA Trans.. 19 (1980) II.
188 E.M. Zacharias and R.A. Parnell, Food Technol., 26 226 CA. Shook and M.C. Roco, Slurry Flow. Butterworth
(1972) 160. Heinemann, 1991. Chapter 10.
189 A. Krigman, In Tech, 29 (1982) 49. 227 M.S.A. Abouelwafa and E.J.M. Kendall. J. Phys. E, 13
190 D.G. Steele, Meas. Control, 14 (1981) 24. (1980) 341.
191 J. Blitz, Fundamentals of Ultrasonics, Plenum, London, 228 T.K. Thiyagarajan, N.S. Dixit. P. Satyamurthy,
2nd edn., 1967, pp. 1266127. N. Venkatramani and V.K. Rohatgi. Meas. Sci. Technol..
192 J.C. Bonnet and L.L. Tavlarides, Ind. Eng. Chem. Res., 2 (1991) 69.
26 (1987) 811. 229 A. Teyssedou, F. Aube and P. Champagne, Meas. Sci.
193 J. Yi and L.L. Tavlarides, Ind. Eng. Chem. Res., 29 Technol.. 3 (1992) 485.
(1990) 475. 230 Y. Jiang and KS. Rezkallah, Meas. Sci. Technol.. 4
194 C. Tsouris and L.L. Tavlarides, Ind. Eng. Chem. Res., 29 (1993) 496.
(1990) 2170. 231 G.F. Hewitt, in G. Hetsroni (Ed.). Handbook of
195 C. Tsouris, L.L. Tavlarides and J.C. Bonnet, Chem. Eng. Multiphase Systems, Hemisphere. Washington. DC. 1982,
Sci., 45 (1990) 3055. Chapter 10.
196 C. Tsouris and L.L. Tavlarides, Ind. Eng. Chem. Res., 32 232 O.C. Jones, in R.J. Goldstein (Ed.). Fluid Mechanics
(1993) 998. Measurements, Hemisphere, Washington. DC, 1983.
197 J.R. Allegra and S.A. Hawley, J. Acoust. Sot. Am., 51 Chapter 10.
(1972) 1545.
198 AS. Ahuja, J. Appl. Phys., 44 (1973) 4863. Viscosity-bused techniques
199 AS. Ahuja, J. Acoust. Sot. Am., 51 (1972) 916. 233 R. Pal, AIChE J., 39 (1993) 1754.
200 R.J. Urick, J. Appl. Phys., 18 (1947) 983. 234 R. Pal, Ind. Eng. Chem. Res.. 32 (1993) 1212.
201 R.J. Urick and W.S. Ament, J. Acoust. Sot. Am., 21 235 R. Pal, Colloids Surfaces. 71 (1993) 173.
(1949) 115. 236 M. Van der Waarden, J. Colloid Sci.. 9 (1954) 215.
202 A.H. Harker and J.A.G. Temple, J. Phys. D, 21 (1988) 237 H. Eilers, Kolloid-Z.. 97 (1941) 313.
1576. 238 J.V. Fitzgerald, F.J. Matusik, H.R. Link and P.C. Scarna.
203 R.A. Williams, C.G. Xie, R. Bragg and W.P.K. M&C Meas. Control, 14 (1980) 150.
Amarasinghe, Colloids Surfaces, 43 (1990) 1. 239 R.T. Albano, M&C Meas. Control, 18 (1984) 202.
204 D.J. Wedlock, C.J. McConaghy and S. Hawksworth, 240 J.V. Fitzgerald and F.J. Matusik. M&C Meas. Control,
Colloids Surfaces, 77 (1993) 49. 20 (1986) 175.
205 A.J. Babchin, R.S. Chow and R.P. Sawatzky, Adv. Colloid 241 J.V. Fitzgerald, F.J. Matusik and T.M. Walsh. M&C
Interface Sci., 30 (1989) 111. Meas. Control, 21 (1987) 118.
206 E.E. Isaacs, H. Huang, A.J. Babchin and R.S. Chow, 242 R. Byrne, M&C Meas. Control. 22 (1988) 281.
Colloids Surfaces, 46 (1990) 177. 243 J.V. Fitzgerald, F.J. Matusik and H.R. Oppliger. M&C
207 E.E. Tsaacs and R.S. Chow, Adv. Chem. Ser., 231 (1991) Meas. Control, 9 (1975) 65.
51-77. 244 M. Jackman, Food Tcchnol.. July ( 1991) 90.
208 T.S.B. Sayers, Colloids Surfaces, 77 (1993) 39. 245 H.A. Barnes, in K. Walters (Ed.), Rheometry: Industrial
Applications, Research Studies Press. Chichester. 1980.
Density-based techniques Chapter 2.
209 G.L. Schlatter and C.E. Miller, US Patent 3 952 592, 1976. 246 Z. Kemblowski, J. Sek and P. Budzynski. Rheol. Acta. 27
210 M.H. November, US Patent 4 055 082, 1977. ( 1988) 82.
211 M.H. November, US Patent 3 906 198, 1975. 247 W. Heinz, Chem. Eng. Commun.. 37 ( 1985) 93,
212 I. Ghahramani, US Patent 4 225 778, 1980. 248 C.W. Brabender Instruments, Convimeter Rotational
213 P.P. Elderton, US Patent 4 059 744, 1977. Process Viscometer for Continuous In-line Viscosity
214 R.E. Talmo, Pasadena and T.H. Lee, US Patent 3 518 Measurements, Bulletin CM80. NJ, 1980.
886, 1970. 249 Haake Inc.-Fison Instruments. The Sofraser In-line
215 P. Taffe, Processing, June (1990) 41. Viscometer for Process Control, Bulletin Vol. 14, no. 1.
216 R.A. Furness, Meas. Control, 20 (1987) 7. NJ, 1989.
R. Pal/Colloids Surfaces A: Physicochem. Eng. Aspects 84 (1994) 141-193 193

250 Bendix Environmental & Process Instruments Division, 252 Warren Spring Laboratory, Report highlights poten-
Bendix Ultra-Viscoson Viscometer, B333SB374, tial of viscometry control, Processing, December (1985)
Lewisburg, WV. 33.
251 C. Martin, Viscosity control for mixed fuel oils, 253 R. Pal, Chem. Eng. J., 43 (1990) 53.
Processing, February (1983) 37.